Using new methods, we measure the intensities of higher-order risk preferences (prudence and temperance) in an incentivized experiment with 658 adolescents. Aligned with theory, we find that higher-order risk preferences are strongly related to field behavior, including prevention, health, addictive behavior, and financial decision-making. Most importantly, we show that ignoring prudence and temperance can yield misleading conclusions about the relation of risk preferences to field behavior, and that survey measures of risk tolerance often relate to field behavior because they capture higher-order risk preferences.