next | table of contents

Executive Summary

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

There is growing awareness that serious, reversible error permeates America's death penalty system, putting innocent lives at risk, heightening the suffering of victims, leaving killers at large, wasting tax dollars, and failing citizens, the courts and the justice system.

Our June 2000 Report shows how often mistakes occur and how serious it is: 68% of all death verdicts imposed and fully reviewed during the 1973-1995 study period were reversed by courts due to serious errors.

Analyses presented for the first time here reveal that 76% of the reversals at the two appeal stages where data are available for study were because defense lawyers had been egregiously incompetent, police and prosecutors had suppressed exculpatory evidence or committed other professional misconduct, jurors had been misinformed about the law, or judges and jurors had been biased. Half of those reversals tainted the verdict finding the defendant guilty of a capital crime as well as the verdict imposing the death penalty. 82% of the cases sent back for retrial at the second appeal phase ended in sentences less than death, including 9% that ended in not guilty verdicts.

Part II of our study addresses two critical questions: Why does our death penalty system make so many mistakes? How can these mistakes be prevented, if at all? Our findings are based on the most comprehensive set of data ever assembled on factors related to capital error-or other trial error.

Our main finding indicates that if we are going to have the death penalty, it should be reserved for the worst of the worst: Heavy and indiscriminate use of the death penalty creates a high risk that mistakes will occur. The more often officials use the death penalty, the wider the range of crimes to which it is applied, and the more it is imposed for offenses that are not highly aggravated, the greater the risk that capital convictions and sentences will be seriously flawed.

Most disturbing of all, we find that the conditions evidently pressuring counties and states to overuse the death penalty and thus increase the risk of unreliability and error include race, politics and poorly performing law enforcement systems. Error also is linked to overburdened and underfunded state courts.


The higher the rate at which a state or county imposes death verdicts, the greater the probability that each death verdict will have to be reversed because of serious error.

  • The overproduction of death penalty verdicts has a powerful effect in increasing the risk of error. Our best analysis predicts that:

    • Capital error rates more than triple when the death-sentencing rate increases from a quarter of the national average to the national average, holding other factors constant.

    • When death sentencing increases from a quarter of the national average to the highest rate for a state in our study, the predicted increase in reversal rates is six-fold-to about 80%.

In particular, the more often states impose death sentences in cases that are not highly aggravated, the higher the risk of serious error.

  • At the federal habeas stage, the probability of reversal grows substantially as the crimes resulting in capital verdicts are less aggravated. For each additional aggravating factor, the probability of reversal drops by about 15%, when other conditions are held constant at their averages. Imposing the death penalty in cases that are not the worst of the worst is a recipe for unreliability and error.

Comparisons of particular counties' and states' capital-sentencing and capital-error rates illustrate the strong relationship between frequent death sentencing and error. For example:

  • Among counties with 600 or more homicides and five or more death sentences during the study period, ten had the highest death-sentencing rates: Pima County (Tucson), Arizona; suburban Baltimore County, Maryland; Clark County (Las Vegas), Nevada; Pinellas County (St. Petersburg), Florida; Oklahoma (City), Oklahoma; Maricopa County (Phoenix), Arizona; Hamilton County (Cincinnati), Ohio; Hillsborough County (Tampa), Florida; Polk County, Florida; and Muscogee County, Georgia. These counties had an average capital error rate of 71% at the first and last appeal stages, and eight of them put a total of 16 people on death row who were later found not guilty. The ten comparable capital counties with the lowest death-sentencing rates are San Francisco, California; Richmond, Virginia; Fulton County (Atlanta), Georgia; Essex County (Newark), New Jersey; St. Louis City, Missouri; Pulaski County (Little Rock), Arkansas; Bernalillo County (Albuquerque), New Mexico; Davidson County (Nashville), Tennessee; Jackson County (Kansas City), Missouri; and Prince George's County (suburban Washington), Maryland. These counties had an average error rate of 41%, and none sentenced anyone to death during the study period or since who was later found not guilty.(Table 16)

  • All but one of the 10 states with the highest death-sentencing rates during the 23-year study period had overall capital reversal rates at or above the average rate of 68%.


Four disturbing conditions are strongly associated with high rates of serious capital error. Their common capacity to pressure officials to use the death penalty aggressively in response to fears about crime and regardless of how weak any particular case for a death verdict is, may explain their relationship to high capital error rates.

  • The closer the homicide risk to whites in a state comes to equaling or surpassing the risk to blacks, the higher the error rate. Other things equal, reversal rates are twice as high where homicides are most heavily concentrated on whites compared to blacks, than where they are the most heavily concentrated on blacks.

  • The higher the proportion of African-Americans in a state-and in one analysis, the more welfare recipients in a state-the higher the rate of serious capital error. Because this effect has to do with traits of the population at large, not those of particular trial participants, it appears to be an indicator of crime fears driven by racial and economic conditions.

  • The lower the rate at which states apprehend, convict and imprison serious criminals, the higher their capital error rates. Predicted capital error rates for states with only 1 prisoner per 100 FBI Index Crimes are about 75%, holding other factors constant. Error rates drop to 36% for states with 4 prisoners per 100 crimes, and to 13% for those with the highest rate of prisoners to crimes. Evidently, officials who do a poor job fighting crime also conduct poor capital investigations and trials. Well-founded doubts about a state's ability to catch criminals may lead officials to extend the death penalty to a wider array of weaker cases-at huge cost in error and delay.

  • The more often and directly state trial judges are subject to popular election, and the more partisan those elections are, the higher the state's rate of serious capital error.


Heavy use of the death penalty causes delay, increases cost, and keeps the system from doing its job. High numbers of death verdicts waiting to be reviewed paralyze appeals. Holding other factors constant, the process of moving capital verdicts from trial to a final result seems to come to a halt in states with more than 20 verdicts under review at one time.

Poor quality trial proceedings increase the risk of serious, reversible error. Poorly funded courts, high capital and non-capital caseloads, and unreliable procedures for finding the facts all increase the chance that serious error will be found. In contrast, high quality, well-funded private lawyers from out of state significantly increase a defendant's chance of showing a federal court that his death verdict is seriously flawed and has to be retried.

Chronic capital error rates have persisted over time. Overall reversal rates were high and fairly steady throughout the second half of the 23-year study period, averaging 60%. When all significant factors are considered, state high courts on direct appeal-where 79% of the 2349 reversals occurred-found significantly more reversible error in recent death verdicts than in verdicts imposed earlier in the study period. Other things equal, direct appeal reversal rates were increasing 9% a year during the study period.

State and federal appeals judges cannot be relied upon to catch all serious trial errors in capital cases. Like trial judges, appeals judges are susceptible to political pressure and make mistakes. And the rules appeals judges use to decide whether errors are serious enough to require death verdicts to be reversed are so strict that egregious errors slip through. We study four illustrative cases in which the courts approved the convictions and death sentences of innocent men despite a full set of appeals.* These case studies show that judges repeatedly recognized that the proceedings were marred by error but affirmed anyway because of stringent rules limiting reversals.


The lower the rate at which a state imposes death sentences-and the more it confines those verdicts to the worst of the worst-the less likely it is that serious error will be found. The fewer death verdicts a state imposes, the less overburdened its capital appeal system is, and the more likely it is to carry out the verdicts it imposes. The more often states succumb to pressures to inflict capital sentences in marginal cases, the higher is the risk of error and delay, the lower is the chance verdicts will be carried out, and the greater is the temptation to approve flawed verdicts on appeal. Among the disturbing sources of pressure to overuse the death penalty are political pressures on elected judges, well-founded doubts about the state's ability to convict serious criminals, and the race of the state's residents and homicide victims.


We employ an array of statistical methods to identify factors that predict where and when death verdicts are more likely to be found to be seriously flawed, and to assure that the analyses are comprehensive, conservative and reliable: We use several statistical methods with different assumptions about the arrangement of capital reversals and reversal rates to ensure that results are driven by relationships in the data, not statistical methods. We analyze reversals at each separate review stage and at all three stages combined. We use multiple regression to analyze the simultaneous effect on reversal rates of important general factors (state, county, year and time trend) and specific conditions that may explain error rates. We examine factors operating at the state, county and case level. And we check for consistency of results across analyses to determine which factors and sets of significant factors are the most robust and warrant the most confidence.


The harms resulting from chronic capital error are costly. Many of its evident causes are not easily addressed head-on (e.g., the complex interaction of a state's racial make-up, its welfare burden and the efficacy of its law enforcement policies). And indirect remedies are unreliable because they demand self-restraint by officials who in the past have succumbed to pressures to extend the death penalty to cases that are not highly aggravated. As a result, some states and counties may conclude that the only answer to chronic capital error is to stop using the death penalty, or to limit it to the very small number of prospective offenses where there is something approaching a social consensus that only the death penalty will do.

In other states and counties, a set of carefully targeted reforms based upon careful study of local conditions might seek to achieve the central goal of limiting the death penalty to "the worst of the worst"—to defendants who can be shown without doubt to have committed an egregiously aggravated murder without extenuating factors. Ten reforms that might help accomplish this goal are:

  • Requiring proof beyond any doubt that the defendant committed the capital crime.

  • Requiring that aggravating factors substantially outweigh mitigating ones before a death sentence may be imposed.

  • Barring the death penalty for defendants with inherently extenuating conditions—mentally retarded persons, juveniles, severely mentally ill defendants.

  • Making life imprisonment without parole an alternative to the death penalty and clearly informing juries of the option.

  • Abolishing judge overrides of jury verdicts imposing life sentences.

  • Using comparative review of murder sentences to identify what counts as "the worst of the worst" in the state, and overturning outlying death verdicts.

  • Basing charging decisions in potentially capital cases on full and informed deliberations.

  • Making all police and prosecution evidence bearing on guilt vs. innocence, and on aggravation vs. mitigation available to the jury at trial.

  • Insulating capital-sentencing and appellate judges from political pressure.

  • Identifying, appointing and compensating capital defense counsel in ways that attract an adequate number of well-qualified lawyers to do the work.


Over decades and across dozens of states, large numbers and proportions of capital verdicts have been reversed because of serious error. The capital system is collapsing under the weight of that error, and the risk of executing the innocent is high. Now that explanations for the problem have been identified and a range of options for responding to it are available, the time is ripe to fix the death penalty, or if it can't be fixed, to end it.

* We study the cases of Lloyd Schlup, Earl Washington, Anthony Porter and Frank Lee Smith. See pp. 25-36.
next | table of contents