Experimental studies have suggested that the biosynthesis of arachidonic acid-derived eicosanoids such as prostaglandin E2 via the cyclo-oxygenase pathway may play a significant role in supporting cell proliferation in human prostate cancer cell lines. However, the aetiological significance of this for clinical prostate cancer has remained unclear. In particular, the potential for prostate cancer chemoprevention using nonsteroidal anti-inflammatory drugs (cyclo-oxygenase inhibitors; NSAIDs) has received little attention. The purpose of our study was to investigate associations between prostate cancer risk and use of NSAIDs. A population-based case-control study was carried out over 13 months from 1996 in metropolitan Auckland, New Zealand. A total of 317 newly diagnosed prostate cancer cases (including 192 "advanced" cases) representative of all cancer cases in the study population were identified from urology clinic referrals and histology reports. A total of 480 age-matched controls were recruited following random selection from the study population using electoral rolls as the sampling frame. After adjusting for potential confounding by socio-economic status and dietary fat consumption, there was a trend toward reduced risks of advanced prostate cancer associated with regular use of total NSAIDs (RR = 0.73; 95% CI 0.50-1.07) and total aspirin (RR = 0.71; 95% CI 0.47-1.08). However, these associations failed to reach statistical significance at the usually accepted levels. Weaker inverse associations were found for total prostate cancers, which included a number of small, low-grade tumours of less clinical significance. These findings lend support to proposed underlying aetiological hypotheses which imply a role for cyclo-oxygenase activity in prostate cancer progression.