The preferences assumed to govern intertemporal trade-offs are generally considered to be stable economic primitives, though evidence on this stability is notably lacking. We present evidence from a large field study conducted over two years, with around 1,400 individuals using incentivized intertemporal choice experiments. Aggregate choice profiles and corresponding estimates of discount parameters are unchanged over the two years and individual correlations through time are high by existing standards. However, some individuals show signs of instability. By linking experimental measures to administrative tax records, we showthat identified instability is uncorrelated with both levels and changes in sociodemographic variables.