Testing and analysis has been performed on square steel thin-walled tubes fabricated using symmetrically-overlapping U-channels and foam. This research analyzes flange-to-flange attachment, effect of foam in the columns, effect of adhesive stiffness, and influence of steel thickness, as related to the local buckling loads, global buckling loads, and crippling loads. Four 14-foot (4.27 m) foam-filled, thin-walled, galvanized steel columns were manufactured by Novatek, Inc. and tested in axial compression with pinned boundary conditions. For three of the four configurations, the two-piece 4-in. (10.2-cm) square shell surrounded prefabricated polystyrene foam inserts; the fourth column had no foam insert. The column outer shells were composed of two 16-gauge galvanized steel channels with overlapping flanges and the webs on opposite sides of the column. The two adjacent flanges on each side of the columns were adhesively bonded together in all cases. In addition to the adhesive, two columns had either periodic screws or short welds spaced evenly along the length of the columns to delay the onset of flange buckling of the outer channel, and potentially increase the compression strength. The other two columns had adhesive only bonding the flanges, one of which had no foam filler. The various configurations all exhibited similar compression strengths. Failure for all columns initiated with local buckling, followed by global buckling and local crippling, which occurred simultaneously. The method of flange attachment, the effect of the foam in the columns, and flange thicknesses were isolated and analyzed using mechanics-based analysis, parametric studies, and finite element analysis. The results show the ideal spacing of screws or short-welds, if used, is less than or equal to 5 in (12.7 cm) for the given column length. This increases the local buckling load to the Euler buckling load and preserves the original shape of the cross-section. The adhesive needs only a tensile strength of approximately 1 ksi (6.4 kPa) to prevent local buckling for any spacing of screws or short-welds, but needs to be applied uniformly (much of the adhesive in the column tests had been scraped off of the flanges during assembly). The results also show that foam core does not increase the Euler buckling load, but does increase the crippling load by delaying inward buckling of the column webs and flanges. Using foam with the given stiffness and a yield strength of 50 psi (345 kPa), uniform foam-to-steel bonding could increase the crippling strength up to 21% even without adhesive between the flanges. Using adhesive with the given stiffness between the flanges could increase the crippling strength by up to 63% without foam. The crippling strength could increase up to 72% if both adhesive between the flanges and a foam insert are used.