next | previous | table of contents

A Broken System, Part II:
Why There is So Much Error in Capital Cases,
and What Can be Done About It

I. Introduction: Rising Doubts About the Administration of the Death Penalty

The nation is in the midst of its most serious and sustained public discussion of the death penalty in decades.1 The discussion has many parts, but one theme is constant: People who have not committed a crime for which the law permits the death penalty should not be sentenced to die.2 Yet, as Supreme Court Justice Sandra Day O'Connor said recently, "If statistics are any indication, the system may well be allowing some innocent defendants to be executed." 3

A. The Recent Wave of Rethinking and Reform

Over the last two years, these concerns have led to changes in the public's thinking about the death penalty and a variety of policy initiatives:

  • The Governor of Illinois declared a moratorium on executions in that state, and appointed a blue ribbon committee to address the problem.4 In the meantime, the Illinois Supreme Court issued comprehensive rules to improve the quality of capital trials in the state,5 and the Illinois House of Representatives approved an even more demanding package of reforms.6

  • The Nebraska Legislature passed a moratorium on executions so that a comprehensive study of the state's death penalty could occur. The moratorium was vetoed, but the study went forward, finding economic and geographic disparities and proposing reforms.7

  • In the Winter-Spring 2001 legislative cycle, 37 of 38 death penalty states (all but Kansas) considered legislation to reform and moderate the use of the death penalty; at least one such reform became law in 21 of those states.8

  • Maryland's Governor also ordered a comprehensive study of that state's death penalty.9 In March 2001, that state's House of Delegates passed a bill imposing a moratorium on executions until the study is completed. The bill reportedly had majority support in the state Senate but was blocked by an end-of-session filibuster.10 Days later, the state's high court imposed its own moratorium, granting review on an issue affecting all Maryland cases.11

  • In Nevada, the state Senate passed legislation by a wide margin imposing a moratorium on executions. The Assembly instead proposed and the Legislature adopted a bill funding a broad study of the state's death penalty.12 The Legislature also directed the study commission to consider broadened availability of DNA tests and bans on executing mentally retarded individuals (such a ban passed the State Assembly13 ) and juvenile offenders.14

  • In Texas:

    1. The Legislature passed and the Governor signed emergency legislation giving death row inmates greater access to DNA evidence that might exonerate them.15

    2. The Legislature adopted and the Governor signed legislation for the first time ensuring timely and non-patronage appointment of, and minimum standards for, defense lawyers, affording extra assistance to lawyers handling capital cases and providing nearly $20 million (compared to zero previously) to help pay for counsel.16

    3. The state also raised the amount it provides innocent people found to have been erroneously convicted from a maximum of $50,000 to $250,000.

    4. Both houses adopted, but the Governor vetoed, legislation banning execution of mentally retarded persons.17

    5. The Governor endorsed and the Senate and a House committee adopted, but the full House defeated, a bill making life without parole the alternative to a death sentence.18

    6. The House passed a bill banning executions for crimes committed by juveniles.

    7. The Senate Criminal Justice Committee unanimously approved a bill barring consideration of the defendant's race in deciding whether to sentence him to die and requiring hearings on the issue in appropriate cases.

    8. Committees in both houses endorsed a resolution letting voters decide whether to impose a 2-year moratorium on executions while the state's death penalty is studied.

    9. The number of people executed dropped from 40 in 2000 to 17 in 2001.19

  • North Carolina adopted a new law allowing prosecutors to seek life without parole, not the death penalty, for aggravated first-degree murder. The absence of this power had previously inflated the number of death sentences in marginal cases.20

  • In Virginia, which has long put strict limits on review of capital verdicts, the Legislature and Governor agreed on "a major overhaul of the state's death-penalty laws, giving those on death row the right to seek new DNA testing."21

  • "Major campaigns to suspend executions have been launched in 19 [eventually 20] states" in 2001,22 including moratorium and death-penalty study bills in Alabama (passed by the Senate Judiciary Committee), Arkansas, Delaware, Illinois, Indiana, Kentucky,23 Louisiana, Maryland (discussed above24 ), Missouri, Mississippi, Nevada (discussed above25 ), New Jersey, North Carolina, Oklahoma, Pennsylvania, Tennessee, Texas (discussed above26 ), Virginia and Washington-up from 2 in 1999 and 2000 combined, and virtually none during the rest of the 1990s. Abolition bills nearly passed in New Hampshire and New Mexico this Spring, and were proposed in Illinois, Indiana, Kentucky, Louisiana, Missouri, Mississippi, Montana (hearings held in House Judiciary Committee), Nebraska, New Jersey, New York, Oregon, Pennsylvania, Tennessee and Virginia.27

  • Bills to reinstate the death penalty failed in Maine and Massachusetts by large margins.28

  • Major death penalty studies have occurred or are occurring in Arizona,29 Connecticut, Illinois, Indiana, Maryland, Nebraska, Nevada, New Jersey, North Carolina, Virginia, and the Federal system.30

  • At least 26 states across the country, including Arizona, Arkansas, California, Delaware, Florida, Idaho, Indiana, Louisiana, Maryland, Missouri, Nebraska,31 New Mexico, North Carolina, Ohio, Oklahoma, Oregon, Pennsylvania, Tennessee, Texas, Utah, Virginia and Washington, have recently adopted reforms making it possible for capital prisoners to obtain post-conviction DNA testing.32 Over half the statutes were passed in 2001. Similar legislation was considered in Alabama, Kentucky, Nevada, New Jersey, Mississippi, North Carolina, Ohio, South Dakota (adopted by legislature but vetoed by the Governor) and Tennessee.33

  • Six states-Arizona, Connecticut, Florida, Missouri, North Carolina and Tennessee- barred the death penalty for mentally retarded persons in 2001, joining 13 states that had done so in the previous two decades.34 Not least among the motivations for such action is that retarded people are especially vulnerable to being convicted and condemned when innocent-as happened to three men exonerated in 1999-2001: Anthony Porter, Earl Washington and Albert Burrell.35 Similar concerns may have led the United States Supreme Court to agree to reconsider a 12-year-old ruling allowing states to execute mentally retarded defendants.36

  • Bans on executing juveniles were considered this year in Arkansas (passed by the Senate), Florida (passed by the House), Indiana, Kentucky (endorsed by the Governor) and South Carolina, in addition to Nevada and Texas, discussed above.37

  • Plans for improving capital representation were adopted in Arkansas, Mississippi, Oklahoma and Virginia,38 in addition to Texas (discussed above39 ). Proposals for similar improvements are under consideration in Alabama, Georgia, North Carolina and Washington.40

  • Bipartisan legislation to assure capital prisoners' access to DNA testing, to improve the quality of representation at capital trials and to increase penalties for prosecutorial suppression of evidence and other causes of unreliable capital verdicts has been proposed in Congress. Its Senate co-sponsors include Gordon Smith (R-Or.), Susan Collins (R-Me.) and John Warner (R-Va.), all death penalty supporters. In the House it has 214 co-sponsors, 44 of them Republican, including respected conservatives John Boehner (OH), Jennifer Dunn (WA), Lindsey Graham (SC), Ray LaHood (IL), George Nethercutt (WA), Rob Portman (OH) and Joe Scarborough (FL).41

  • Prominent death penalty supporters have recently expressed doubts about the reliability of, and in some cases the continuing need for, the nation's death penalty, including: Illinois Republican Governor George Ryan,42 Connecticut Republican Governor John Rowland (citing concerns about unfair use of the death penalty against racial minorities),43 New Mexico Republican Governor Gary Johnson44 and Minnesota Reform Party Governor Jesse Ventura;45 conservative commentators Oliver North,46 George Will (Washington Post),47 Bruce Fein48 and Paul Craig Roberts49 (both in the Washington Times), Rod Dreher (New York Post),50 Gerald Seib (Wall St. Journal)51 and Emmett Tyrell, Jr. (American Spectator);52 the CIA's former station chief in Pakistan (expressing concerns about the effect of the death penalty on America's ability to extradite and try terrorists);53 Timothy McVeigh prosecutors James Orenstein (citing concerns about race discrimination)54 and Beth Wilkinson;55 conservative religious leaders Rev. Pat Robertson,56 John Diulio (director of President Bush's faith-based public service initiative),57 The Christian Life Commission of the General Board of Texas Baptists,58 the North Carolina Conference of The United Methodist Church,59 Rutherford Institute founder and president John W. Whitehead60 and the editors of the influential evangelical Christian magazine Christianity Today;61 the Republican author of Ohio's death penalty law (now a Justice on its Supreme Court),62 the author of Texas's death penalty statute (now a senior state district judge),63 the just-retired Chief Justice of the Washington Supreme Court and his successor64 and other conservative officials in a variety of states who have supported proposals for death penalty moratoria or other reforms.65

  • Oklahoma Republican Governor Frank Keating, speaking to the National Press Club, and John Podhoretz-both staunch death penalty supporters-recently called for laws limiting death verdicts to cases where guilt is proved to an "actual certainty" and not just "beyond a reasonable doubt."66

  • A leading legal academic defender of the death penalty recently wrote that "[w]e are witnessing today a true crisis of confidence in the death penalty in the United States. For the first time in more than twenty-five years, public support for the death penalty seems to be waning. The evidence of trouble is everywhere."67

  • Supreme Court Justice Ruth Bader Ginsburg recently took the unprecedented step of endorsing a death penalty moratorium, noting that she has "yet to see a death penalty case among the dozens coming to the Supreme Court on eve-of-execution stay applications in which the defendant was well-represented at trial."68

  • Not long afterwards, Justice Sandra Day O'Connor, a long-time death penalty supporter who helped write Arizona's death penalty statute before moving to the nation's high court, gave speeches in two states raising "'serious questions' about whether the death penalty is being fairly administered in the United States" and whether "the system may . . . be allowing some innocent defendants to be executed." Justice O'Connor said the "quality of defense lawyers for people in capital cases has been inadequate in too many cases."69

  • Since 1994, support for the death penalty in national polls has dropped from about 80% to a 20-year low of 63%.70 When the choice posed is between life without parole and the death penalty (life without parole is in fact the alternative to death sentences in all capital states except Texas71 ) support for the penalty drops to 50% or below.72 Half of all Americans support a suspension of executions while the death penalty is studied.73

  • Changes in public opinion may be having an important effect on public action. Jurors and judges are imposing fewer death sentences, which dropped 29% between 1998 and 2000 (the last year when data are available).74 The number of executions declined 32% between 1999 and 2001.75 Governors are commuting more death sentences.76

B. Error and the Risk of Executing the Innocent as the Spur to Reform

Catalyzing national concern about the reliability of capital convictions and sentences is the fact that, in the same year the nation carried out its 700th execution in the modern death-sentencing era, it also approached its 100th exoneration of a person sentenced to die despite being factually or legally innocent.77

The capital system is supposed to single out offenders who committed crimes for which the law allows the death penalty, and to carry out the penalty with the swiftness and sureness needed to provide credible retribution and deterrence. The statistics to which Justice O'Connor and others refer-99 exonerations nationally compared to about 750 executions78 -are a good test of the quality of the outcomes the capital system produces. One alarming failure for every 7 or 8 routine successes is a troubling bottom line. This would be true for any process. It is especially true for one run by the Government that says who lives and who dies and takes years, even decades, to fix its mistakes.

As Justice O'Connor further notes, there is an even more disturbing lesson from the ratio of exonerations to executions: When so many failures are known to occur, and when mistakes are so hard to catch, it is likely that some are not caught. There is reason to fear that some executions counted as successes are actually undiscovered failures—executions of defendants who were innocent, or did not commit a crime for which the death penalty is allowed, but whom the courts inadvertently allowed to be executed.79

Regrettably, researchers cannot study or accurately count the number of innocent people executed in the United States since 1973, because available information often cannot show whether executed people were innocent. The truth may die with them or with the victims they were convicted of killing. The death penalty is unique in this respect. Unlike other situations in which innocent people are accidentally killed-an airplane crash, an e-coli outbreak or an explosion at a chemical plant-the total number of innocent fatalities caused by the death penalty is unknowable, even when all evidence is examined.

Nor can researchers examine all the evidence. Along with the press, policy makers and the public, researchers are often denied access to the single existing repository of evidence, such as DNA samples, that can best establish whether executed people were innocent: The police file in the case of executed persons. Notwithstanding requests from the media, churches, charities and the relatives of executed individuals, and even in cases involving weak and compromised evidence of guilt and repeated protestations of innocence, officials routinely deny access to police files on executed individuals,80 or even destroy the files.81 The reason officials give for withholding and destroying files increases concern about what the files would show-that if the evidence were subjected to DNA and other tests, and if the tests went badly for the state, "'it would be shouted from the rooftops that the Commonwealth of Virginia [or another state] executed an innocent man.'"82

Nor do American capital jurisdictions conduct inquests like those used in Britain, Canada and Australia to review questionable criminal convictions and sentences after judicial review has ended and sentences have been carried out.83 The one state that is currently conducting such an inquest due to concerns about faulty testimony by state forensic experts has decided that questionable "cases in which inmates have already been executed will be considered last, if at all."84 These obstacles to research and informed policy making are especially disturbing where people may have been executed though they were innocent of a crime. They are also troubling where people may have been sentenced to death for crimes they did commit, but for which death is not a legal penalty.

The death penalty thus is unique among government actions (or private ones, for that matter) that put innocent life at risk. Only in this context are researchers and policy makers unable to learn how many innocent casualties an activity causes—including because officials decline to disclose evidence on the question.85 For the same reason, the death penalty is uniquely an area where researchers and policy makers must rely entirely on evidence of a risk that the state is taking innocent lives, i.e., on the amount and rates of near misses and other serious errors found before the worst harms occurred.

Consideration of risk is of course appropriate in regard to all potentially harmful activities, even where investigators can identify innocent fatalities once they occur, so harms can be prevented. Delta Airlines, Hormel Meats or Goodyear Tires would risk severe heavy fines and loss of business if they stopped conducting safety inspections or ignored the results until it was shown that their products and services had injured or killed people. If risk management based on flaws observed in quality control inspections is a crucial way to assess the safety and effectiveness of enterprises in which actual injuries and fatalities can be detected,86 it is indispensable where unavoidable circumstances and the withholding of information in government files keep the public from even knowing whether innocent fatalities have occurred.

C. The Disturbing Amount of Error, Unreliability and Risk Revealed by our June 2000 Report, A Broken System

Our June 2000 report, A Broken System: Error Rates in Capital Cases, 1973-1995 ("A Broken System"), documented the rates of serious, judicially reversible error that the American capital system's quality control inspectors discovered between 1973 and 1995. Based on results from all three levels of judicial review of death verdicts-direct appeals to state high courts, state post-conviction review by state trial and high courts and federal habeas corpus review—A Broken System made the following findings, among others in the National Report Card, p. 9 below:

  • State high courts found that 41% of the over 4500 capital verdicts finally reviewed on direct appeal in the 23-year study period were too flawed to carry out and had to be reversed.87

  • Of the 59% of capital verdicts that survived direct review, at least 10% were found to be flawed and were reversed by state post-conviction courts.88

  • Of the remaining 54% of death verdicts, 40% were reversed by federal habeas judges because of serious error.

  • Overall, only 32% of the capital verdicts imposed and finally reviewed between 1973 and 1995 were approved for execution.* Conversely, 68% of those verdicts were so seriously flawed that they had to be reversed and sent back for a new trial of guilt or sentence.

National Composite Capital Punishment Report Card


First Death Sentence


First Direct Appeal


First Consensual Execution


First Non-Consensual Execution


Years From First Death Sentence to First Non-Consensual Execution


Average Number of Years from Sentence to Execution


Average Number of Years from Sentence to Final Federal Relief


Sentences and Executions

Total Number of Death Sentences


Total Number of Executions


Percentage of Death Sentences Carried Out


Error Rates

State Direct Appeal (34 States)

Number Reviewed on Direct Appeal


Number Reversed on Direct Appeal


Percentage Reversed on Direct Appeal


Number Awaiting Direct Appeal


Percentage Awaiting Direct Appeal


Number Forward to State Post-Conviction


State Post-Conviction (28 States)

Number Reviewed on Post-Conviction


Number Reversed on Post-Conviction


Percentage Reversed on Post-Conviction


Number Forward to Federal Habeas Corpus


State Direct Appeal and State Post-Conviction Combined (28 States)

Overall Rate of Error Found by State Courts


Federal Habeas Corpus (28 States)

Number Reviewed on Habeas


Number Reversed on Habeas


Percentage Reversed on Habeas


Overall Rates Including [and Excluding] State Post-Conviction (28 States)

Overall Error Rate

68% [65%]

Overall Success Rate

32% [35%]

Sources: DRCen; Death Row U.S.A., Fall 2001; DADB; HCDB; Appendix C

For reasons discussed above,89 A Broken System could not and did not study the number of people executed for crimes they did not commit or for crimes for which the law does not allow the death penalty. What it instead studies is the number and rates of seriously flawed, and thus unreliable, death verdicts, and the resulting risk that people have been, and will continue to be, executed for crimes they did not commit or ones for which the law does not allow the death penalty.

The next part of this Report explains why the high number of flawed capital verdicts poses an intolerable risk of seriously unreliable capital outcomes, including execution of the innocent, and why it therefore is imperative to study the causes of high rates of capital error. Thereafter, Parts III-VII identify and interpret factors associated with high error rates and with error itself at the state level (where overall capital policy is largely made), the county level (where decisions whether to charge capitally and whether to convict and condemn are mostly made), and in individual federal habeas cases. Part VIII then lays out some options for reducing the risk of serious error and thus of executing innocent people and others for whom the law bars death as a punishment.

* The proportion surviving direct review was 59%: 100% - 41% reversed on direct appeal = 59%. The proportion surviving direct appeal and state post-conviction was 53%: 59% surviving direct review - 10% of those 59% that were reversed on state post-conviction = 59% - 6% = 53%. The proportion surviving all three stages was 32%: 53% surviving the first two stages - 40% of those 53% that were reversed on federal habeas corpus = 53% - (.40 x 53%) = 53% - 21% = 32%. (Note: The numbers in the National Report Card below are calculated without rounding at each stage of the calculation, but only at the end. For ease of exposition, the example in this footnote rounds at each stage of the calculation.)

next | previous | table of contents