Intoxicated drivers from a 10-weekend roadside survey of vehicles randomly selected from the streets of a medium-sized southeastern metropolitan area are compared to officially arrested drivers apprehended during a corresponding time period. Upon examination, no statistically significant arrest bias was detected by the driver's race. However, compared to the roadside survey population, lower occupational status drivers (i.e., blue collar) were significantly more likely to appear in the arrested population than an upper status (i.e., white collar) driver, especially in the lightly patrolled middle and upper class residential sections of town. The significance of occupational status, when combined with the finding that race was not a significant indicator of difference between the arrested and roadside survey populations, suggests that proactive police arrest bias is more a function of institutionally determined police patrol practices rather than overt racial prejudice on the part of individual officers. (Author abstract)