Identifying individuals with recent HIV infection is critical to research related to viral reservoirs, outbreak investigations and intervention applications. A multi-assay algorithm (MAA) for recency of infection was used in conjunction with self-reported date of infection and documented date of diagnosis to estimate the number of participants recently infected in the Strategic Timing of AntiRetroviral Treatment (START) trial. We tested samples for three groups of participants from START using a MAA: (1) 167 individuals who reported being infected /= 2 years before randomization. Participants with low titer and avidity and a baseline viral load > 400 copies/mL were classified as recently infected. A significantly higher percentage of participants who self-reported being infected /= 2 years (65% [109/167] vs. 2.5% [5/199], p < 0.001). Among the 771 individuals who did not know their duration of infection at randomization, 206 (26.7%) were classified as recently infected. Among those diagnosed with HIV in the 6 months prior to enrollment, the 373 participants who reported recent infection (n = 167) or who had confirmed recent infection by the MAA (n = 206) differed significantly on a number of baseline characteristics from those who had an unknown date of infection and were not confirmed by the MAA (n = 565). Participants recently infected by self-report and/or MAA were younger, more likely to be Asian, less likely to be black, less likely to be heterosexual, more likely to be enrolled at sites in the U.S., Europe or Australia, and have higher HIV RNA levels. There was good agreement between self-report of recency of infection and the MAA. We estimate that 373 participants enrolled in START were infected within 6 months of randomization. Compared to those not recently infected, these participants were younger, had higher HIV RNA levels and were more likely to come from high income countries and from populations such as MSM with more regular HIV testing.