International Comparison: Capital Punishment in the US and Pakistan in 2025

In 2025, the United States exe­cut­ed 47 peo­ple, the high­est total in over fif­teen years, while pub­lic sup­port for the death penal­ty fell to a fifty-year low and juries reject­ed death sen­tences in a major­i­ty of cap­i­tal tri­als. That same year, Pakistan held 3,394 peo­ple on death row — one of the largest con­demned pop­u­la­tions in the world — while main­tain­ing a de fac­to mora­to­ri­um on exe­cu­tions that has been in place since December 2019 and fail­ing to imple­ment legal pro­tec­tions its own Supreme Court has mandated.

A new study by Hamza Saeed, a research and com­mu­ni­ca­tions fel­low at DPI and for­mer com­mu­ni­ca­tions offi­cer at Justice Project Pakistan, pro­vides a com­par­a­tive overview of how cap­i­tal pun­ish­ment oper­at­ed in the United States and Pakistan in 2025. It eval­u­ates sen­tenc­ing and exe­cu­tions, vul­ner­a­bil­i­ty among the con­demned, imple­men­ta­tion of legal safe­guards, clemen­cy and last-resort review, wrong­ful-con­vic­tion risk, inter­na­tion­al oblig­a­tions, and pol­i­cy direc­tion against the stan­dards each sys­tem has estab­lished for itself. The analy­sis draws pri­mar­i­ly on the Death Penalty Information Center’s 2025 annu­al report on the United States and the Justice Project Pakistan and National Commission for Human Rights’ 2025 report on Pakistan, with lim­it­ed sup­port­ing evi­dence from the Death Penalty in India Annual Statistics Report 2025.

The United States and Pakistan were select­ed as con­trast­ing reten­tion­ist sys­tems with dif­fer­ent con­sti­tu­tion­al, insti­tu­tion­al and polit­i­cal con­texts. The com­par­i­son exam­ines whether these dif­fer­ent arrange­ments pro­duce sim­i­lar gaps between for­mal safe­guards and actu­al prac­tice. It iden­ti­fies com­mon pat­terns with­out claim­ing that cap­i­tal pun­ish­ment alone caus­es every documented failure.

Key Findings By Dimension

Who is con­demned. Vulnerability is preva­lent among those sub­ject­ed to cap­i­tal pun­ish­ment in both sys­tems. In the United States, 83% of the 47 peo­ple exe­cut­ed in 2025 had at least one doc­u­ment­ed vul­ner­a­bil­i­ty, includ­ing seri­ous men­tal ill­ness, intel­lec­tu­al dis­abil­i­ty indi­ca­tors, brain dam­age or sig­nif­i­cant child­hood trau­ma. In Pakistan, indi­gence, inad­e­quate cap­i­tal-defence rep­re­sen­ta­tion, undi­ag­nosed men­tal ill­ness and, in blas­phe­my cas­es, reli­gious-minor­i­ty sta­tus can com­pound defen­dants’ dis­ad­van­tage. These con­di­tions do not all con­sti­tute cat­e­gor­i­cal legal bars to exe­cu­tion, but they may pro­vide impor­tant mit­i­ga­tion and, in defined cir­cum­stances, engage con­sti­tu­tion­al or legal protections.

The imple­men­ta­tion gap. In both coun­tries, for­mal safe­guards fre­quent­ly pro­vide less pro­tec­tion in prac­tice than their legal word­ing promis­es. In the United States, intel­lec­tu­al-dis­abil­i­ty and com­pe­ten­cy pro­tec­tions may be nar­rowed through rigid def­i­n­i­tion­al thresh­olds, pro­ce­dur­al bar­ri­ers and restric­tions on the retroac­tive appli­ca­tion of lat­er deci­sions. In Pakistan, the Supreme Court has recog­nised pro­tec­tions for pris­on­ers whose severe men­tal ill­ness or dis­abil­i­ty pre­vents them from com­pre­hend­ing the ratio­nale and pur­pose of their pun­ish­ment, but the required assess­ment pro­ce­dures, spe­cial­ist Medical Boards and foren­sic men­tal-health facil­i­ties remain inad­e­quate­ly imple­ment­ed. India illus­trates the wider dif­fi­cul­ty of trans­lat­ing appel­late safe­guards into tri­al-lev­el prac­tice: 79 of 83 analysable cap­i­tal cas­es in 2025 failed to com­ply with the sen­tenc­ing require­ments estab­lished in Manoj, which the Supreme Court sub­se­quent­ly rein­forced in Dupare.

Clemency and last-resort review. Clemency is for­mal­ly avail­able in both coun­tries but pro­vides lim­it­ed and uneven prac­ti­cal pro­tec­tion. In the United States, exec­u­tive clemen­cy peti­tions con­nect­ed with 2025 exe­cu­tions were over­whelm­ing­ly denied, while the Supreme Court sep­a­rate­ly declined to stay any exe­cu­tion dur­ing the year. In Pakistan, 171 mer­cy peti­tions were pend­ing in Khyber Pakhtunkhwa with­out a clear admin­is­tra­tive timetable, pub­lished eval­u­a­tion pro­ce­dure or effec­tive over­sight mech­a­nism. Abdul Basit’s pres­i­den­tial com­mu­ta­tion demon­strates that clemen­cy can oper­ate, but the exten­sive insti­tu­tion­al coor­di­na­tion required in a sin­gle excep­tion­al case also illus­trates the absence of a pre­dictable and accessible process.

International law. The two coun­tries raise dif­fer­ent legal and pol­i­cy con­cerns and should not be assessed as though they have accept­ed iden­ti­cal oblig­a­tions. Pakistan remains bound by the ICCPR and retains cap­i­tal offences, includ­ing non-lethal offences, that are dif­fi­cult to rec­on­cile with the ​“most seri­ous crimes” stan­dard endorsed by its own Supreme Court. The United States remains bound by the ICCPR sub­ject to its reser­va­tions, but has not rat­i­fied the Convention on the Rights of Persons with Disabilities and is not legal­ly bound by UN General Assembly res­o­lu­tions call­ing for a mora­to­ri­um. Nevertheless, its oppo­si­tion to inter­na­tion­al mora­to­ri­um ini­tia­tives and efforts to expand death eli­gi­bil­i­ty indi­cate grow­ing diver­gence from the broad­er inter­na­tion­al trend towards restriction.

Innocence. The irrev­o­ca­bil­i­ty of exe­cu­tion gives par­tic­u­lar sig­nif­i­cance to weak­ness­es in evi­dence, dis­clo­sure and legal rep­re­sen­ta­tion. The United States has record­ed 202 death-row exon­er­a­tions since 1973, while cas­es involv­ing Richard Glossip, Justin Wolfe and Robert Roberson con­tin­ued to demon­strate the impor­tance of post-con­vic­tion review in 2025. Pakistan does not main­tain a com­pa­ra­ble nation­al death-row exon­er­a­tion dataset, mak­ing the scale of wrong­ful cap­i­tal con­vic­tions dif­fi­cult to mea­sure. The absence of com­pa­ra­ble data should not be inter­pret­ed as evi­dence that the under­ly­ing risk is absent.