I measure the effects of street-level political advertising on voting behavior. I use a novel dataset on ad location in a major Spanish city during elections for the national parliament as well as granular socio-economic and voting data. This set-up, where more than two parties are running for office and elections are very competitive, allows me to explore the heterogeneous effects of ads across parties as well as how parties’ ads affect other parties’ vote shares. To identify the effects of parties’ ads, I exploit a legally mandated randomized assignment of ad location to parties across multiple years. I find that a party’s own ads have a positive effect on its vote share, although the effects are heterogeneous across parties. A one standard deviation increase in the number of ads increases a party’s vote share by 0.87 percentage points on average. Ads of parties with ideologically distant platforms consistently have a negative effect on a party’s vote share. In contrast, ads of parties that are close competitors may act either as complements or substitutes in different years.