OBJECTIVES: Interval censoring typically arises when the occurrence of an event is observed according to a pre-defined visit schedule and not at the actual occurrence time. Survival analysis of such data may therefore require caution and Kaplan Meier analysis may provide biased estimation of median time to such events. The objective of this work is to evaluate through an illustrative example how interval censoring can impact and may be handled in the context of time to progression analysis. METHODS: This work compares various approaches to analyse interval-censored data for the estimation of median time to event and hazard ratios: 1) Standard Kaplan Meier analysis 2) Standard Cox and Weibull regression analyses 3) An ad hoc Bayesian Weibull model explicitly modelling the censoring process Methods were compared on a real-world case of a phase 3 trial comparing Vascular Targeted Photodynamic therapy (VTP) vs. Active Surveillance (AS) in early-stage prostate cancer patients followed up over two years. Median time to progression (PFS, in months) and Hazards Ratios were then evaluated and compared based on 2 biopsies at months 12 and 24. RESULTS: Results showed that interval-censored Kaplan Meier analysis could systematically under-estimate PFS, leading up to 50% under-estimation in both arms. However HR estimates showed to be very consistent across the methods used. CONCLUSIONS: Kaplan Meier estimates of median time to progression with interval censoring may be systematically under estimated. However Hazard Ratios estimates are less affected by interval censoring.