Terrestrial laser scanning is becoming increasingly widely used in engineering surveying, especially in the area of displacement monitoring and performing diagnostic measurements. Empirical assessment of laser scanner accuracy in real measurement conditions is of great importance when it comes to assessing the suitability of scanners for specific engineering applications. This paper describes an empirical analysis of the accuracy of two pulsed laser scanners: Leica ScanStation C10 and Riegl VZ-400i. The tests were conducted in two measurement environments: inside and outside a building, where test bases were designed and made in the form of a set of several reference points, i.e. reflective targets. During the tests, the deviations of spatial distances, spatial angles and coordinates were analysed on the basis of the results of scanner measurements in relation to precise total station measurements treated as reference. Additionally, the accuracy of the reference sphere approximation was tested. The results obtained indicate that the values of spatial distance deviations between pairs of targets for measurements performed inside the building are comparable for both scanners. The mean absolute error (MAE) was 1.6 mm for the Leica scanner and 1.9 mm for the Riegl scanner. For measurements outside the building, the Leica scanner proved to be more accurate. The MAE of spatial angle deviations between targets for measurements inside the building was 54cc for the Leica scanner and 96cc for the Riegl scanner, while outside the building, it was 29cc for the Leica scanner and 73cc for the Riegl scanner. The average error in matching the local scanner system with the reference coordinates inside the building was 0.9 mm for the Leica scanner and 1.3 mm for the Riegl scanner, and outside, it was 0.8 mm for the Leica scanner and 2.5 mm for the Riegl scanner. Additionally, the accuracy of the reference sphere approximation was tested. The analyses carried out prove that even an older scanner model (ScanStation C10) has high measurement accuracy useful in engineering surveying despite the significantly slower measurement compared to modern scanner models.