OBJECTIVE: A short pregnancy interval (PI) has been associated with increased child mortality, but mechanisms are unclear. We studied factors associated with PI and the effect of PI on birthweight and haemoglobin. METHODS: Information was analysed from 2218 multigravidae who were recruited at the prenatal clinic (1758) or in the labour ward (460) of the Provincial Hospital in Kisumu between June 1996 and July 2000 for a study to assess the interaction between placental malaria and vertical HIV transmission. RESULTS: The HIV prevalence was 28.9%. HIV seropositivity, older age, being unmarried, and <8 years of education were associated with a prolonged PI; among all women, a stillbirth, abortion, or death of a liveborn child as outcome of the previous pregnancy, and death of a child other than the last born among HIV-seronegative women, were associated with a shortened PI. No significant effect of short PI (an interval <24 months) on low birth weight (LBW), prematurity, small-for-gestational-age infants or maternal anaemia was evident. An abortion, stillbirth, or death of a liveborn child as outcome of the previous pregnancy was associated at the present delivery with LBW among HIV-seronegative women [adjusted odds ratio (AOR) 3.33, 95% confidence interval (CI) 1.63-6.81], and a low haemoglobin (<11 g/dl) among HIV-seropositive women (AOR 2.01, 95% CI 1.05-4.03 in the third trimester). CONCLUSION: Public health efforts to ensure 'adequate' birth spacing may run contrary to family planning decisions to replace a deceased child and may be spent on prenatal issues like prevention of anaemia, and vertical HIV transmission.