Particle image velocimetry (PIV) measurements in polydisperse sprays have been investigated wing Monte Carlo simulations. The objective of the study was to determine how using droplets with a wide size distribution for the "seed" particles in PIV measurements affects the mean velocity obtained from the cross-correlation of the image pairs. Simulations were conducted using a variety of droplet size distributions, including a uniform distribution (in which the probability density function is a constant value for all allowable droplet sizes), log-normal size distributions, Rosin-Rammler size distributions, and a size distribution determined experimentally using phase Doppler interferometry. It was found that polydispersity has a strong effect on the measured velocity field, with the measured mean velocities being weighted toward the velocities of the larger droplets. The estimated mean velocity resulting from the cross-correlation algorithm was found to be approximately equal to the mean velocity weighted by the droplet surface area for all of the size distributions investigated. Signal saturation in the images was also found to affect the cross-correlation of the image pairs and the resulting mean velocities.