Aim To examine the use of crystal methamphetamine ( crystal meth) and its association with high-risk sexual behaviour among gay men in London. Design Cross-sectional surveys using self-administered questionnaires. Settings National Health Service ( NHS) out-patient human immunodeficiency virus (HIV) treatment clinic in London; NHS HIV testing/sexual health clinics in London; central London gyms. Participants HIV-positive gay men attending the HIV treatment clinic in 2002-03 (n = 388); HIV-negative gay men attending the HIV testing/sexual health clinics in 2002-03 ( n = 266); gay men using the gyms between January and March 2003 ( n = 445), 2004 ( n = 653) and 2005 ( n = 494). Measurements Percentage of gay men who reported: (i) using crystal meth in the previous 12 months; (ii) using other recreational drugs ( e. g. cocaine, ecstasy, ketamine); (iii) high-risk sexual behaviour in the previous 3 months. Findings The percentage of gay men who had used crystal meth in the previous year varied by sample ( HIV treatment clinic, 12.6%; HIV testing/sexual health clinics, 8.3%; gyms, 19.5%; P < 0.001) but did not change over time in the gyms ( 19.8%, 20.7%, 17.8%; P = 0.5). In all samples, the majority of men used crystal meth only once or twice a year. Most crystal meth users (> 80%) had taken other recreational drugs. Crystal meth and other drug users were more likely to report high-risk sexual behaviour than other men, e. g. HIV treatment clinic sample: crystal meth users, 34.7%; other drug users, 18.9%, non-users, 10.6%, P < 0.001. Cause and effect could not be established. Conclusion Among gay men in London surveyed in clinics, approximately one in 10 reported using crystal meth in the previous 12 months (HIV-positive men 12.6%; HIV-negative men 8.3%). Most men used it infrequently only once or twice a year. There was no evidence of increasing use of crystal meth between 2003 and 2005.