Estimation of design hourly volume (DHV), commonly the 30th highest hourly volume (30HV) in a year, from sample counts is an important aspect in traffic engineering practice. Directional design hourly volume (DDHV) is usually obtained by multiplying a parameter of directional split to DHV. The normally used methodology to estimate DHV is developing a consistent and predictable relationship between the annual average daily traffic (AADT) and the DHV. However, highway designers have questioned this methods validity and its limitations have been well discussed. Due to the fact that recreational travel on most holidays in developed countries is very active and leads to vast increases in highway traffic volumes. The main objective of this paper is to develop more accurate and efficient DDHV prediction models based on directional hourly volumes occurring in holiday periods. In this paper, the conventional method is reviewed first. Then, holiday traffic peaking characteristics are investigated based on past 20 years of data from permanent traffic counters on primary rural highways in Alberta, Canada. With the recognition of holiday traffic peaking phenomena, genetic algorithms (GAs) are employed to assist in developing a number of Victoria Day based DDHV prediction models corresponding to different types of roads. At the end of this paper, discussions regarding these models are presented.