Fitting probabilistic models to data is often difficult, due to the general\nintractability of the partition function. We propose a new parameter fitting\nmethod, Minimum Probability Flow (MPF), which is applicable to any parametric\nmodel. We demonstrate parameter estimation using MPF in two cases: a continuous\nstate space model, and an Ising spin glass. In the latter case it outperforms\ncurrent techniques by at least an order of magnitude in convergence time with\nlower error in the recovered coupling parameters.\n