Objective To investigate the effect of interpregnancy interval (IPI) on preterm birth (PTB) according to whether the previous birth was preterm or term. Design Cohort study. Setting USA (California), Australia, Finland, Norway (1980–2017). Population Women who gave birth to first and second (n = 3 213 855) singleton livebirths. Methods Odds ratios (ORs) for PTB according to IPIs were modelled using logistic regression with prognostic score stratification for potential confounders. Within‐site ORs were pooled by random effects meta‐analysis. Outcome measure PTB (gestational age <37 weeks). Results Absolute risk of PTB for each IPI was 3–6% after a previous term birth and 17–22% after previous PTB. ORs for PTB differed between previous term and preterm births in all countries (P‐for‐interaction ≤ 0.001). For women with a previous term birth, pooled ORs were increased for IPI <6 months (OR 1.50, 95% CI 1.43–1.58); 6–11 months (OR 1.10, 95% CI 1.04–1.16); 24–59 months (OR 1.16, 95% CI 1.13–1.18); and ≥ 60 months (OR 1.72, 95%CI 1.60–1.86), compared with 18–23 months. For previous PTB, ORs were increased for <6 months (OR 1.30, 95% CI 1.18–1.42) and ≥60 months (OR 1.29, 95% CI 1.17–1.42), but were less than ORs among women with a previous term birth (P < 0.05). Conclusions Associations between IPI and PTB are modified by whether or not the previous pregnancy was preterm. ORs for short and long IPIs were higher among women with a previous term birth than a previous PTB, which for short IPI is consistent with the maternal depletion hypothesis. Given the high risk of recurrence and assuming a causal association between IPI and PTB, IPI remains a potentially modifiable risk factor for women with previous PTB. Tweetable Short versus long interpregnancy intervals associated with higher ORs for preterm birth (PTB) after a previous PTB. Tweetable Short versus long interpregnancy intervals associated with higher ORs for preterm birth (PTB) after a previous PTB.