We expand the best-of-n model for mate choice by including cost of searching and cost of courtship. Computer simulations show that increasing costs strongly reduce the optimal number of males to be sampled. If the female is already familiar with a male, this would also reduce the number of males sampled and bias the choice in favour of this particular male. In a field test we removed 271 female Pied Flycatchers Ficedula hypoleuca soon after mating and released them 3-4 km away. The density of unmated males in the study area was high at any one time, and searching costs seemed to be low. If mate fidelity is caused by high searching costs we would expect a low degree of mate fidelity in this experiment. However, most returning females reunited with their previous mate, despite having failed in their first nesting attempt with him. Among those females that returned to the area, faithful females did not renest more quickly than unfaithful females, and females did not renest more quickly if they had been mated longer before removal. This indicates that the cost of courtship was low and could not explain the observed mate fidelity. Instead, we suggest that the mate fidelity was due to competition between females for a mate, or to the fact that the mates of faithful females were of higher than average quality.