The maximum flexural strength of I-sections depends on its cross-section classification. A cross-section is classified as plastic, compact, semi-compact or slender depending on the classification of its web and flange plates. These section classification limits in different international codes are recommended based on certain simplistic assumptions. For example, in classifying the web as slender (class IV) or semi-compact (class III), Eurocode assumes that the web is simply-supported at the flange edges, while AISC and AASHTO assume nearly fixed boundary conditions. The proposed research looks at re-evaluating the web classifications at the limit of slender sections, by means of eigen value buckling analyses on a wide range of practical I-sections. The research shows that the flange provides varying degrees of fixity to the web plate, ranging from simply-supported to fixed conditions, depending on its size relative to the web. Based on these findings, a new web section classification is proposed. The work further shows that this section classification not only improves the strength predictions of class III sections that are currently conservatively classified as class IV sections, but also improves the flexural strength predictions of class IV sections, by allowing greater effective widths of the cross-sections. The section classification also determines the stress distribution in the cross-section at its ultimate limit state. For instance, design provisions recommend the use of full plastification of plastic and compact sections, while limiting the extreme flange fiber stress to its yield stress in a semi-compact section. The extreme flange stress is limited to the yield stress in a slender section, while only considering the effective cross-section after local buckling. However, the ultimate stress distributions in hybrid girders, (with lower web yield strengths, resulting in early web yielding as compared to the flange) is unclear. This work is extended to understand the behaviour of hybrid girders, with particular emphasis on slender-web hybrid sections. With the aid of full geometric and material non-linear analyses, it is observed that the maximum flexural strength of a hybrid girder is attained when the web is partially plastified, accompanied by the redistribution of stresses from the web to the flange. A stress distribution model for hybrid girders at their ultimate strength is proposed, by estimating the partial depth of plastification in the compression region of the web, as a function of the flange and web dimensions.