We investigate models for uplink interference in wireless systems. Our models account for the effects of outage probabilities. Such an accounting requires a nonlinear, even nonconvex model, since increasing interference at the receiving base station increases both mobile transmit power and outage probability, and this results in a complex interaction. Our system model always has at least one solution, a fixed point, and it is provably unique under certain reasonable conditions. Our main purpose is to model real wireless systems as accurately as possible, and so we test our models on realistic scenarios using data from a sophisticated simulator. Our algorithm for finding a fixed point works very well on such scenarios, and is guaranteed to find the fixed point when we can prove it is unique. A slightly simplified model reduces the main data structure for a K-sector market to 16K 2 bytes of memory.