Objectives: In a prospective, comparative effectiveness study, we assessed clinical and psychological factors associated with switching from active surveillance (AS) to active treatment (AT) among low-risk prostate cancer (PCa) patients.Methods: Using ultra-rapid case identification, we conducted pretreatment telephone interviews (N = 1139) with low-risk patients (PSA ≤ 10, Gleason≤6) and follow-up interviews 6-10 months post-diagnosis (N = 1057). Among men remaining on AS for at least 12 months (N = 601), we compared those who continued on AS (N = 515) versus men who underwent delayed AT (N = 86) between 13 and 24 months, using Cox proportional hazards models.Results: Delayed AT was predicted by time dependent PSA levels (≥10 vs. <10; HR = 5.6, 95% CI 2.4-13.1) and Gleason scores (≥7 vs. ≤6; adjusted HR = 20.2, 95% CI 12.2-33.4). Further, delayed AT was more likely among men whose urologist initially recommended AT (HR = 2.13, 95% CI 1.07-4.22), for whom tumour removal was very important (HR = 2.18, 95% CI 1.35-3.52), and who reported greater worry about not detecting disease progression early (HR = 1.67, 1.05-2.65). In exploratory analyses, 31% (27/86) switched to AT without evidence of progression, while 4.7% (24/515) remained on AS with evidence of progression.Conclusions: After adjusting for clinical evidence of disease progression over the first year post-diagnosis, we found that urologists' initial treatment recommendation and patients' early treatment preferences and concerns about AS each independently predicted undergoing delayed AT during the second year post-diagnosis. These findings, along with almost one-half undergoing delayed AT without evidence of progression, suggest the need for greater decision support to remain on AS when it is clinically indicated.