My dissertation seeks to understand how reciprocity affects transaction outcomes and mechanism design in online markets. The first essay examines negative reciprocity illustrated as feedback-revoking behavior in the eBay auction market, focusing on its impact and implications for reputation system design. I utilize the biggest policy change of eBay's reputation system in its history as a natural experiment setting to infer the causal impact of the reputation system on seller behavior. I find that strategic engagement in negative reciprocity enables low quality sellers to manipulate their reputations and masquerade as high-quality sellers. I further show that these sellers react strongly to eBay's announcement of a ban on revoking. Interestingly, disallowing negative reciprocity motivates these sellers to significantly improve their service quality. The second essay examines positive reciprocity in one of the leading online barter markets for books, focusing on participants' different reciprocity strategies and their impacts on transaction outcomes. I find that, whereas market participants who use the immediate reciprocity strategy are able to motivate higher service quality for the current transaction from the other partner, participants who use the delayed reciprocity strategy derive more benefits for future transactions by fulfilling their wishlists sooner. I further show that the market participants can be segmented into different reciprocity strategies based on their book avidness, breadth of interest, and psychographic profiles. Overall, the two studies provide important theoretical and practical implications for the design and regulation of online markets.