Individuals frequently face intertemporal decisions. For the purposes of economic analysis, the preference parameters assumed to govern these decisions are generally considered to be stable economic primitives. However, evidence on the stability of time preferences is notably lacking. In a large field study conducted over two years with about 1,400 individuals, time preferences are elicited using incentivized choice experiments. The aggregate distributions of discount factors and the proportion of present-biased individuals are found to be unchanged over the two years. At the individual level, the one year correlations in measured time preference parameters are found to be high by existing standards, though some individuals change their intertemporal choices potentially indicating unstable preferences. By linking time preference measures to tax return data, we show that identified instability is uncorrelated with socio-demographics and changes to income, future liquidity, employment and family composition.