Loss aversion has been shown to be an important driver of people’s investment decisions. Encouraged by regulators, financial institutions are in search of ways to incorporate clients’ loss aversion in their risk classifications. The most critical obstacle appears to be the lack of a valid measurement method for loss aversion that can be straightforwardly incorporated into existing processes. This paper presents the results of two large-scale implementations of such a method within a risk-profiling application of an established financial institution. In total, we elicit loss aversion for 1,040 employees and 3,740 clients. We find that the observed distributions align with existing findings, and that loss aversion is largely independent of the risk-return preferences commonly used for investor classification. Furthermore, the correlations we observe between these two preferences and individuals’ background characteristics align with those observed in the literature. Loss aversion is strongly related to education—higher educated individuals being more loss averse—whereas risk aversion is strongly related to gender, age, and clients’ financial situation—women, more senior, and less wealthy participants being more risk averse. These findings support the conjecture that risk and loss aversion are complementary in capturing investor preferences.