This paper examines dynamic user equilibrium in a bi-modal corridor with risk-averse travelers. Travelers make choices between a risky (stochastic) and congested highway against a reliable (deterministic) and crowded public transit line. A mean-variance approach is adopted to measure the travel cost under risk. At user equilibrium state, nobody can reduce his/her travel cost by ultimately changing his/her departure time or mode choice. We derive traffic flow distribution and congestion evolution patterns at user equilibrium state. It is found that three departure time intervals may exist in the sequence of time on the highway, i.e., intervals in which travelers always arrive early, may be early or late, or always arrive late at the destination. The middle interval becomes longer when travelers are more risk-averse. Unlike the risk-neutral case where the departure rate of highway traffic decreases with departure time, the departure rate of highway traffic may first decrease and then increase with departure time when travelers are highly risk-averse. In addition, when travelers are more risk-averse, the travel cost becomes higher and the public transit ridership also increases. Finally, we prove that perfect traffic information provision is welfare-improving for risk-averse travelers.