Traffic load models available in building codes are most often developed for short or medium span bridges, but most unfavorable traffic situations for long span bridges are very different from the ones considered in them. For this reason funds may be used irrationally, if inappropriate traffic load models are used for long span bridge design. Weigh – in – Motion (WIM) data from WIM station installed on 72. kilometer of highway A1, have been used in these thesis. First data cleaning was performed, then data were split into two lanes. Long span bridge loads were calculated by using information about vehicles found in traffic flow from the cleaned WIM data. Load calculations were done for 200, 300, 400, 500 and 600 meter long spans, traffic flow was simulated using seven different traffic scenarios, out of which first six simulates traffic with varying percentage of trucks, the seventh scenario simulates traffic flow consisting entirely out of cars. For each lane, span, traffic scenario combination Gumbel’s distributions were fitted to the highest 30% of the calculated loads, by using maximum likelihood estimate for left truncated data; loads were extrapolated to the probability of exceedance of 10% in hundred years period. Results show that Eurocode 1 part 2 load model 1 loads are too conservative for use in long span bridge design even when very unlikely scenario of only trucks in the leftmost lane is considered.