This paper presents an experimental investigation about the shear lag effects of tension steel members with welded connections. In total twenty tests were conducted, including twelve on single angles and eight on single tees. Various weld arrangements were considered as test parameters for the angle specimens which were connected by their short legs. For the tees, the influences of weld length and flange-to-web width ratio were studied. The test results showed that, the test efficiency, which is defined as the ratio of the ultimate capacity to the calculated tensile capacity of the gross section, can be increased when a balanced weld arrangement instead of an unbalanced one was employed for the angle specimens. For the tee specimens, the test efficiency was not sensitive to the weld length. In addition, increasing the flange-to-web width ratio could increase the test efficiency of the tees, although the improvement was not significant. Subsequently, design specifications for shear lag were examined, and the test-to-predicted ratios were presented. For more extensive comparisons, the ratios from an earlier investigation undertaken by the authors and co-workers and those from other researchers were also included for discussion. Finally, with the available database, reliability studies were undertaken to examine the resistance factors used in the current codes for predicting the ultimate tensile capacity of single angles and tee sections, and accordingly, new design recommendations are proposed.