The work presents the effect of the B-type guiderail joints of the SP-05/2 barrier, of the N2-W4-A class, on the course of the TB11 and TB32 virtual crash tests. The guiderail was modelled as a continuous beam (without joints) and as a system consisting of 4 m long segments connected by beam elements reflecting screw joints. The TB11 test concerns a passenger car with a mass of 900 kg, impacting onto barrier with a velocity of 100 km/h, at an angle of 20°. The TB32 test concerns a passenger car with a mass of 1500 kg, onto barrier with a velocity of 110 km/h, at an angle of 20°. The numerical calculations were done using the LS-Dyna finite element code, with the use of Geo Metro and Dodge Neon car models taken from the National Crash Analysis Center (NCAC) public library. The car models were modified to the relevant extent. The results of the virtual crash tests were analyzed with respect to the parameters and acceptance criteria of crash tests, as required by PN-EN 1317-1:2010 and PN-EN 1317-2:2010 standards. Taking the guiderail segment joints into consideration enables checking whether the guiderail will maintain continuity during a crash test. Modelling crash tests with guiderails modelled as a continuous beams results in decreasing nearly all crash parameters, and in case of the TB11 test – in smoother repulse of the car.