Purpose u The purpose of this paper is to contribute to a better understanding of the impact of market liquidity on the daily tracking error of exchange-traded funds (ETFs). It puts a special focus on the liquidity cost of individual underlying stocks as well as the process of creation/redemption of ETF shares as key determinants of tracking ability. Design/methodology/approach u The study is based on daily observations of fund data for eight fully replicating German equity ETFs for July 2001-October 2013. A regression model with fund fixed effects is chosen to determine the effect of liquidity cost, creation/redemption and other control variables on daily tracking error. Data were compiled from issuer websites and Datastream. Proprietary XETRA Liquidity Measure, which was used as proxy for liquidity cost was supplied by Deutsche Borse. Findings u The study finds daily tracking error to significantly depend on the liquidity of underlying stocks. This finding emerges even though the ETFs in this study predominantly use in-kind creation/redemption. Even after controlling for creation/redemption, the liquidity impact remains basically unchanged. One reason might be imperfect replication of index weights: Either the in-kind-basket delivered in the course of creation/redemption does not perfectly match the benchmark-weights or the internal rebalancing of weights causes liquidity cost. Originality/value u To the best of the authorsE knowledge, this is the first paper that uses a specific liquidity measure for each single stock underlying an ETF. The findings extend the literature by corroborating the view that liquidity of individual stocks in the underlying portfolio has an impact on tracking error.