A seasonal variation in the presentation of Pneumocystis jirovecii pneumonia (PCP) has been reported and a previous study from this centre noted a seasonal variation in mortality rates. This study examined seasonal influences (including climatic factors) within-host factors (clinical and laboratory-derived variables), the infectious burden of P. jirovecii in bronchoalveolar lavage (BAL) fluid, the presence of dihydropteroate synthase (DHPS) mutations in P. jirovecii, variations in knowledge and skills of junior medical staff, and mortality in 547 episodes of PCP occurring in 494 HIV-infected patients. The overall mortality rate was 13.5%. There was a seasonal variation in mortality: highest in autumn (21.2%) and lowest in spring (9.7%), P = 0.047. After adjustment was made for prognostic factors previously identified as being associated with mortality (increasing patient age, second/third episode of PCP, low haemoglobin, low PaO2, presence of medical co-morbidity and pulmonary Kaposi sarcoma), there was no seasonal association with mortality, P = 0.249. The quantity of P. jirovecii DNA in BAL fluid showed no evidence of seasonal variation, P = 0.67; DHPS mutations were identified with equal frequency in each season and the mortality rate for February and August (when junior medical staff arrive in new posts) was 16.7%, only slightly greater than for other months (13.0%).