Background:
Men who have sex with men (MSM) continue to comprise most new HIV infections in the United States.

Objective:
To measure trends in HIV prevalence and sexual risk behaviors in two community-based probability samples of MSM conducted in 2003 and 2006.

Method:
MSM participated in two random-digit dial telephone surveys conducted in Seattle in 2003 (n=400) and 2006 (n=400). Behavioral measures were based on the preceding 12 months. We adjusted all estimates according to the 2003 age and income distributions.

Result:
Thirteen percent of men in 2003 and 12% of men in 2006 reported being HIV-positive. Similar proportions of men reported anal sex (61% in both surveys) and unprotected anal intercourse with a partner of unknown or discordant HIV status (non-concordant UAI; 12% and 9%, respectively, P = 0.28). The proportion of men reporting 100% condom use decreased slightly (25% vs. 20%, P = 0.14), while the proportion of men reporting unprotected anal intercourse only with partners of the same HIV status (concordant UAI) rose from 24% in 2003 to 32% in 2006 (P = 0.05). More men in 2006 than in 2003 reported meeting a sex partner via the Internet (17% vs. 26%, P = 0.03), while the proportions who met partners through other venues were unchanged. The percentage of sexually active HIV-negative MSM who tested for HIV in the preceding 12 months declined from 74% to 63% (P = 0.03).

Conclusion:
Between 2003 and 2006, the proportion of MSM reporting non-concordant UAI remained stable, while the proportion who reported concordant UAI rose modestly. Fewer sexually active HIV-negative MSM reported recent HIV testing.

Implications:
MSM may be more frequently serosorting while fewer MSM are consistently testing for HIV. Health agencies should develop messages related to serosorting, promote frequent HIV testing, and monitor HIV testing frequency as a prevention indicator.