This paper examines search with active learning and correlated information. We first develop a simple model to show how correlation affects the decision to acquire information. A unique data set on fishing site choice by mid-Atlantic clam fishermen is used to test the model predictions. Results find that clam fishermen search new sites when the catch at familiar sites declines, i.e., when the opportunity cost of gathering information is low, and also when catch at familiar sites is on the rise. Search following a catch decline occurs at spatially distant sites whereas search following a catch increase occurs at nearby sites. Correlated learning is crucial for explaining the site choice patterns in our data. These results provide new insights that may extend to a variety of economic search problems where correlated learning is important.