As the second largest cryptocurrency by market capitalization and today's biggest decentralized platform that runs smart contracts, Ethereum has received much attention from both industry and academia. Nevertheless, there exist very few studies about the security of its mining strategies, especially from the selfish mining perspective. In this paper, we aim to fill this research gap by analyzing selfish mining in Ethereum and understanding its potential threat. First, we introduce a 2-dimensional Markov process to model the behavior of a selfish mining strategy inspired by a Bitcoin mining strategy proposed by Eyal and Sirer. Second, we derive the stationary distribution of our Markov model and compute long-term average mining rewards. This allows us to determine the threshold of computational power that makes selfish mining profitable in Ethereum. We find that this threshold is lower than that in Bitcoin mining (which is 25% as discovered by Eyal and Sirer), suggesting that Ethereum is more vulnerable to selfish mining than Bitcoin.