For the next few decades, cars with autonomous capabilities will still require drivers to resume control of the vehicle, when necessary, known as “takeovers”. However, takeovers are often difficult for drivers as they typically arise in time-sensitive situations. Two experiments were conducted to test the effectiveness of takeover warning designs. The first experiment explored whether two-stage warnings led to better takeover performance than single-stage warnings. Thirty-six undergraduate students completed two drives in autonomous mode on a 2D driving simulator where they took over control of the vehicle when alerted by a takeover warning. The second experiment investigated how pre-warning location and content impacted driver takeover performance. Forty undergraduate students completed four autonomous drives on a driving simulator. In both experiments, initial deceleration, distance to the hazard at the first brake pedal input, and distance to the hazard at the end response constituted takeover performance. Additionally, participant feedback about AV features uncovers potential drawbacks in warning designs that may be of interest to human factors experts. Findings from Experiment 1 show that that two-stage warnings led to better takeover performance than single-stage warnings. In Experiment 2, infotainment-based pre-warnings and dynamic pre-warnings led to prompter takeovers and higher helpfulness ratings than dash-based pre-warnings and static pre-warnings. This research provides stakeholders with takeover alert designs that enable quicker responses and safer stopping distances.