Participants had a mean age of 63.5 years, were on average 5.3 years past diagnosis, and were treated with surgery (59%), radiation (23%), surgery and radiation (11%) or other treatments (7%). The mean ICIQ score at baseline was 6.6(4.9) and only 16 (4%) participants reported no urinary incontinence during the study. Unconditional models showed ICIQ scores varied at baseline (β=6.44, p<.0001), but did not change over two years (β=-.009, p=.295). Treatment group assignment did not predict severity (β=-.38, p=.372) or change in ICIQ scores (β=-.009, p=.293). Age, race, and prostate cancer stage were not associated with severity or change in ICIQ scores. In univariate models, the following characteristics were associated with ICIQ severity, but not change in ICIQ scores over time: obesity (β=1.63, p=.003 ), health related quality of life (β=-.08, p<.0001), comorbidities (β=.44, p=.004), use of anti-incontinence medications (β=1.94, p=.007), time since prostate cancer diagnosis (β=.09, p=.045), EPIC bowel symptoms (β=-.08, p<.0001), EPIC sexual symptoms (β=-.05, p<.0001), climacturia (β=3.33, p<.0001), and adherence to pelvic floor muscles exercises (p<.0001). The type of prostate cancer treatment predicted the severity of ICIQ scores in the following rank order: radiation and surgery (β=1.65, p=.013), surgery only (p<.0001), radiation only (β=-2.04, p<.0001), other treatments (β=-3.14, p<.0001). The type of urinary incontinence predicted severity of ICIQ scores in the following rank order: mixed urinary incontinence (β=8.3, p<.0001), stress urinary incontinence (β=6.61, p<.0001), insensible or continuous urinary incontinence (β=5.06, p<.0001), urgency urinary incontinence (β=4.28, p<.0001), and post-void urinary incontinence (β=4.1, p<.0001). Table 1 presents the multivariate model which showed ICIQ scores were more severe in the control group and were independently associated with obesity, health related quality of life, adherence to pelvic floor muscles exercises, the type of urinary incontinence, and climacturia. Figure 1 illustrates differences between the treatment and control groups in the ICIQ scores predicted by the multivariate model.