A new algorithm for hybrid state estimation, the K-Limited Mode-Change (KLMC) algorithm, is presented. Given noisy measurements, this algorithm estimates the continuous and discrete state histories for a class of hybrid systems that exhibit limited mode changes over time. The KLMC algorithm is compared to an existing hybrid state estimator, the Interacting Multiple Model (IMM), using a newly developed performance metric based on the concept of probability of error. Monte Carlo methods are used to obtain numerical estimates of the performance metric for simple hybrid system models. Simulation results show that KLMC outperforms IMM in terms of the estimate-error metric but requires larger storage and computational resource consumption.