In early phase oncology dose finding studies, the goal is to identify the maximum tolerated dose that has the potential to be efficacious for a single agent or a combination of two or more agents. We discuss the toxicity probability interval (TPI) approach (Ji et al. 2007) and the calibration-free modified TPI approach (Ji et al. 2010), which are adaptive designs that allow for the development of decision rules for toxicity intervals using a Bayesian framework. These types of designs are appealing to non-statisticians because they are easy to implement and dosing decisions can be pre-specified in the protocol. The statistical and practical considerations for the TPI/mTPI including choice of target toxicity rate, evaluation window, and MTD estimation will be presented along with examples illustrating the methods for both single agent and combination trials.