Ramp metering has proven to be an effective freeway management strategy and has been widely implemented throughout the U.S.; however, to date there is no specific document that provides performance data on vehicle acceleration for the design of acceleration lane length at metered ramps. At present, the AASHTO Green Book acceleration lane length design guidance is currently employed by most state DOTs for new proposed ramp metering sites or retrofitting of existing unmetered ramps. Nevertheless, it was found that the acceleration length recommendations documented in the Green Book have not been updated since the 1950s; those outdated data may not be suitable for the modern vehicles and driver behavior. In this regard, this dissertation aimed to investigate the actual acceleration characteristics at metered ramps with various geometric features, and also to figure out the speed profiles to guide the design length of acceleration lanes at metered ramps. Vehicle location versus time information was collected via parallel cameras placed at seven metered ramps in California; a piecewise-constant acceleration model, which assumes vehicles made uniformly accelerated motion within each short space or time interval, was employed to model driver acceleration characteristics. The percentile distance versus speed profiles at each ramp were developed, and regression models were generated to predict the required acceleration length at a given merge speed. The 85th percentile data were recommended as the minimum acceleration length to accommodate the majority of drivers accelerating to a safe merging speed. The results of acceleration characteristics studies showed that acceleration rate at metered ramps is not a constant; drivers tended to accelerate at a higher acceleration rate when speed was lower and vice versa. It was found that length of an existing acceleration lane was the primary factor that affecting driver acceleration behavior; drivers at ramps with a short existing acceleration lane tended to accelerate at higher acceleration rates; this also indicated that ramp metering affects drivers’ perception of acceleration lane length, thus impacting driver acceleration behavior. Lastly, this research proved that assuming constant acceleration did not accurately reproduce the realistic acceleration profiles, and thus, cannot be directly used for determining the required acceleration lengths at various merge speeds. With consideration of the unique operational requirements at metered ramps, this dissertation recommends that the design of acceleration lane length be based on dual-level acceleration lane length design recommendation. The conservative design recommendation is appropriate for ramps that have sufficient space (both existing and proposed metered ramps); while the aggressive design recommendation could be used for existing metered ramps that have insufficient ramp space, or have recurrent ramp queue spillover issues. The recommended acceleration lengths were compared with the AASHTO Green Book design guidance; results indicated that the Green Book design guidance could be reduced by 10 percent (conservative design recommendations) to 35 percent (aggressive design recommendations) for passenger cars at metered ramps. Lastly, this study demonstrated that the required acceleration length for trucks is up to 1.6 times greater than the acceleration lane length design guidance provided in the Green Book.