Our tests: What we looked for and found

We tested convertible car seats in three key areas: how they performed in simulated crash tests, how easy they were to use, and how securely each model fit in five vehicle types.

We performed our crash protection tests according to the federal standard at an outside laboratory, with the exception of some tests in which we installed and crash-tested the seats using a 3-point (lap-and-shoulder) seatbelt. (The federal standard only calls for installation using a 2-point lap-belt.) Since most current vehicles are equipped with 3-point lap-and-shoulder belt systems, we tested the car seats using them, as well. (See box The federal crash standard: Realistic to today's vehicles?)

The crash protection test evaluates the effect a simulated frontal crash has on the car seat and the child-sized dummy in the seat. It simulates a 30 mph crash with a rigid barrier, such as a wall.

All the seats we tested passed the federal crash protection tests, and most performed very well in terms of crash protection. Differences in overall score and differentiation among seats are based mostly on our ease-of-use and fit-to-vehicle evaluations.

Ease-of-use was evaluated by Certified Child Passenger Safety Technicians (such as the installers at a car-seat clinic). They checked—in part—how heavy the seat was; the clarity and usefulness of the labels and instructions; how easy it was to adjust all the straps, and use the clips and buckles; and whether it had and how easy it was to read the level indicators used to get the correct recline angle when rear-facing.

Our fit-to-vehicle tests evaluated how easy it was to correctly install the seats in the rear seats of five different vehicles: a small car (2008 Chevrolet Aveo), family sedan (2008 Honda Accord), small SUV (2008 Jeep Liberty), minivan (2008 Chrysler Town & Country), and mid-sized SUV (2009 Honda Pilot). All installations were made rear- and forward-facing by Certified Child Passenger Safety Technicians.

See the Ratings (available to subscribers) for how the seats performed in various test areas.

We also checked each seat for its ability to fit a newborn test dummy, since the lower-weight range for almost all the seats was five pounds. We checked each seat's harness position and tension, and crotch strap position relative to the newborn dummy using any infant inserts or pillows provided with the seat. (See the recommended models (available to subscribers) for those that provided the best newborn fit).