This paper provides field experimental evidence on the prevalence and determinants of discrimination and in-group favoritism in trust decisions. We observe choices of about 1,000 inhabitants of the city of Zurich who take part in a sequential trust game, in which first movers can condition their investments on the residential districts of second movers. Our main results can be summarized as follows: First movers discriminate significantly in their investment choices, i.e., strangers receive different investments depending on the district they live in. The systematics of the discrimination pattern is underlined by data from an additional newspaper study, where participants correctly guessed the outcome of the study. In terms of district characteristics two factors seem to be key for a district's reputation: while expected trustworthiness of a district increases in the socio-economic status it decreases in the degree of ethnic heterogeneity. Observed discrimination is not just based on mistaken stereotypes but can at least partly be classified as statistical discrimination. This can be inferred from the fact that, on a district level, both expected return on investment and actual investments are positively correlated with actual back transfers. First movers correctly anticipate different levels of trustworthiness and discriminate accordingly. Furthermore, we provide evidence of in-group favoritism, i.e., people trust strangers from their own district significantly more than strangers from other districts. Finally, we discuss individual determinants of discrimination and in-group favoritism.