The impact of transparency on the extent of reciprocal behavior is investigated in a simple repeated gift exchange experiment, where principals set wages and agents respond by choosing effort levels. In addition to the efforts the principals? payoffs are determined by a random component. It is shown that direct reciprocal behavior is much stronger in a more transparent situation where efforts are revealed to the principals. However, there is no significant impact of transparency on average effort as non-transparency leads to a stronger diversity in behavioral patterns such that at the same time the frequency of very low and very large effort levels increases.