In the present work, results of tests conducted on thin welded rectangular steel-concrete stubs are presented. The studied section was made of two cold steel plates with U shape and welded (with electric arc) to form a steel box. The cross section dimensions were:  . The main studied parameters were: the height (50, 100, 150, 200, 300, 400, 500 mm), the effect of the in filled concrete and its age, the discontinuous weld. The tests were carried out at 28 days and 3 years after the date of casting. All tests were achieved under axial compression in a 50 tf machine up to failure. A total of 21 stubs were tested, 8 were empty, 8 filled with concrete whose gravel was made of crushed crystallized slag tested at 28 days of casting and 8 composites as the previous but tested after 3 years. The aim of the study is to bring some light on the behaviour of such composite section. Also, to provide some evidence that the use of crushed slag could be integrated in the manufacturing of non-conventional concrete. All failure loads were predicted numerically and by using the Eurocodes EC3 and EC4 from test results it was confirmed that the length of empty stubs had a drastic effect on the load carrying capacity and the failure mode was rather a local buckling mode with steel sides deformed outwards and inwards. Both numerical EC3 predictions were higher and on the unsafe side when compared to experimental corresponding loads for empty steel samples. For composite stubs, the load carrying capacity increased significantly; the EC4 numerical load predictions were higher in the higher range 300–500 mm and lower in the higher range 50–200 mm. The failure mode of composite stubs was a local buckling mode with all steel sides deformed outwards. The experimental loads obtained after 3 years of casting were higher than the corresponding tested at 28 days. The load ratio (3 years/28 days) was found to be increasing linearly with the increase of the stubs height. More test results are needed to check the EC3 and EC4 validity.