The application of horizontal and hydraulically fractured wells for producing oil from low permeability formations has changed the face of the North American oil industry. One feature of the production profile of many such wells is a transition from transient linear oil flow to boundary-dominated flow. The identification of the time of this transition is important for the calibration of models that forecast the wellÕs future production and the expected ultimate recovery. It is preferable that such models generally use data from the boundary-dominated flow regime for parameter calibration. Accurate well production forecasts are needed for operational decisions, long-term planning, commercial transactions, regulatory proceedings, and asset valuation. Petroleum engineers frequently make the call on the transition point based on subjective visual interpretations of log-log plots for individual wells. This is time-consuming and is generally not repeatable by other analysts. This note evaluates statistical approaches that can serve as alternatives to the subjective visual interpretations. Specifically, the predictive performance of production models calibrated with boundary-dominated data based on transition dates calculated with constrained nonlinear least squares and Bayesian regressions was very close to that obtained using the visual method, suggesting that statistical approaches may indeed be constructed to replace less objective visual approaches without loss of accuracy.