This paper studies a labor market where workers have incomplete information about the quality of their employment match. The model allows past experience to provide information about the quality of a new match. Allowing workers to learn from past job experience generates a decline in job finding and job separation rates with age that is consistent with patterns found in the data. To provide evidence of this learning mechanism, the model generates a prediction that wage volatility on a new job should decline with past job experience. This decline in wage volatility is documented in data from NLSY79.