If values are ubiquitous in science, then we can no longer use the presence of values to discriminate between good and bad science. Some scientific hypotheses can be empirically well-supported and value-laden. How? Much depends on the nature of empirical support, and the definition of values. I argue that values can function as empirical claims, and that where relevant and well-supported by evidence, values can increase the empirical strength of particular scientific theories. Responding to recent commentaries, I analyze the empirical strength of feminist values in particular and show that the evidence-based nature of these values is neither a weakness nor an idealization.