The article investigates the influence of joints on the warping torsion of cold-formed steel bars. The modern warping torsion theory suggests cold-formed steel bars to be simply supported or fixed at the ends. Simple support provides zero warping restraint. Fixed support provides fool warping restraint at the joint of the bar. In real constructions cold-formed steel joints are partial warping restrained. Not considering the partial restraint of deplanations by real joints leads to an incorrect assessment of the twist angles and the stress state of thin-walled steel bars in warping torsion. This article deals with an experimental and analytical investigation of warping torsion of cold-formed steel bars with bolted joints. Considered 142C16, 142C20, 262C23 and 262C29 sections. Five types of joints considered: a wall and both flanges of the bar end sections are fixed; the upper and lower flanges are fixed; the wall is fixed; the wall and the lower flange are fixed; the lower flange is fixed. First, analytical expressions for twist angles and bimoments for warping torsion for bars with partial warping restraints obtained. Analytical results are compared with the results of the warping torsion experiment conducted at Moscow State University of Civil Engineering. The cold-formed steel specification is shown to be a poor predictor for the twist angle and bimoment value of twisting members. The warping factor coefficient is recommended for the estimation of the degree of the joint warping constraint. Experimental values of warping factors for different joint types are obtained. The influence of partial warping restraints and cross-section deformation on the work of the tested cold-formed steel bars are evaluated.