The default requirement in drug development programs is that two or more confirmatory studies must be successful to gain regulatory approval of an investigational treatment. Typically, a sponsor will estimate the probability of technical success (PTS) of a drug development program using a crude approach, often using a qualitative estimation approach rather than a quantitative approach informed with results from previous research. Even when a quantitative approach is used, the PTS of a program is estimated assuming underlying study level parameters are independent leading to an overly conservative estimation. In this work we propose a Bayesian latent relationship modeling (BLRM) framework that explicitly incorporates interdependencies of study level parameters and then predicts ‘success’ using joint statistical modeling and simulations. This novel BLRM is a holistic approach resulting in sensible estimates of PTS without being overly conservative to inform drug development decisions. We illustrate the methodology with a few phase III programs with different, endpoints, and dependency structures.