Pakistan recorded a sharp month-on-month fall in militant attacks during October 2026, but the same dataset contains a warning that a headline decline can hide: kidnappings rose to their highest monthly level in a year. For residents, businesses and people travelling in affected districts, the difference matters. An attack count measures incidents; a kidnapping count measures people taken. Neither figure alone describes the full security situation.
The Pakistan Institute for Conflict and Security Studies (PICSS), an Islamabad-based independent research organisation, counted 143 militant attacks in October, compared with 226 in September. The decline reversed two months of rising attacks. It is a meaningful improvement in the monthly count, but it is not evidence that the wider security challenge has ended.
What changed between September and October?
| Indicator | September 2026 | October 2026 | Change |
|---|---|---|---|
| Militant attacks | 226 | 143 | Down 37% |
| Deaths caused by attacks | 260 | 111 | Down 57% |
| People injured in attacks | 242 | 69 | Down 71% |
| Suicide attacks | 3 | 1 | Down by 2 incidents |
| People kidnapped in militant incidents | 72 | 88 | Up 22% |
The first three measures moved in the same direction. Attacks fell by 83 incidents, deaths from attacks fell by 149, and injuries fell by 173. The single suicide attack recorded in October caused no reported deaths or injuries, unlike the three suicide attacks in September, which caused 45 deaths and 138 injuries.
There is an important distinction in the casualty figures. The 111 deaths refer to people killed in militant attacks. PICSS separately counted 272 deaths when security forces’ counterterrorism actions are included. That wider figure comprised 53 security personnel, five peace committee members, 45 civilians and 169 militants. Combining these categories without explaining the difference would produce a misleading comparison.
The regional picture is not uniform
Balochistan recorded 71 attacks in October, down from 132 in September. Deaths caused by attacks in the province fell from 136 to 44, while injuries declined from 63 to 37. The scale of that reduction made Balochistan the largest contributor to the national month-on-month improvement.
Mainland Khyber Pakhtunkhwa, excluding the merged districts, recorded 40 attacks compared with 58 in September. Deaths from attacks fell from 90 to 38 and injuries from 165 to 19. The merged districts recorded 30 attacks, against 34 a month earlier; deaths fell from 33 to 25, while injuries increased slightly from 11 to 13.
Punjab recorded two attacks, which caused four deaths. PICSS recorded no militant attacks in Sindh, Islamabad, Azad Jammu and Kashmir or Gilgit-Baltistan during the month. A zero in a monthly dataset means no incident was recorded under that dataset’s definition and period; it should not be interpreted as proof that a region has no security risks.
Why the kidnapping figure changes the interpretation
PICSS counted 88 people kidnapped in incidents linked to militant violence in October, compared with 72 in September. It was the highest monthly kidnapping figure in the preceding 12 months. Balochistan accounted for 70 of the 88 people, close to four-fifths of the national total.
This is the most important counter-signal in the report. A decline in attacks does not necessarily mean every form of militant activity is declining at the same pace. Kidnapping creates a different burden for families, communities, local authorities and businesses. It can disrupt travel and work even when the number of explosive or armed incidents falls.
The dataset does not establish one motive for every abduction, and it would be wrong to assign a single explanation to all cases. The defensible conclusion is narrower: the number of people kidnapped increased while attacks, deaths and injuries decreased.
One quieter month does not erase the year-to-date trend
For the first ten months of 2026, PICSS recorded 1,376 militant attacks. Attacks and security forces’ actions together resulted in 3,902 deaths and 1,953 injuries over the same period. Compared with the corresponding period of 2025, the number of attacks was 53% higher and the combined death toll was 36% higher.
This is why October should be described as a better month than September, not as proof of a sustained national turnaround. The monthly comparison and the year-to-date comparison answer different questions. The first shows what changed recently; the second shows how the year is progressing against the previous year.
How readers should use these figures
- Check the location: national totals do not tell a resident whether a particular road, district or neighbourhood is safe today.
- Keep attack and operation figures separate: the number of people killed in attacks is not the same as the total killed in attacks and security operations.
- Watch kidnappings as a separate indicator: October’s increase was concentrated heavily in Balochistan.
- Look for more than one month of movement: a sustained change requires several comparable reporting periods.
- Follow local advisories: use current instructions from district administrations, police and other competent authorities for travel or public events.
Pakistan’s security debate also has a regional dimension. Verity Pakistan has separately examined the Pakistan–Afghanistan security dispute and the question of verifiable action. That context should not be used to assume that every incident in the October dataset has the same cause.
What to watch in the next report
The next meaningful test is whether attacks remain below the September level while kidnappings fall from October’s 88. The provincial breakdown will matter as much as the national total, particularly whether Balochistan’s kidnapping count changes. Until several months of comparable data are available, the most accurate reading is mixed: October brought a substantial fall in attacks and casualties, but a 12-month high in kidnappings, against a year-to-date picture that remains worse than 2025.
Original data release: PICSS October 2026 security statistics.