Strut-and-tie modelling constitutes a powerful tool for the design of complex structural reinforced concrete elements. It has been proved numerically that strut-and-tie (ST) models obtained using structural optimization methods produce designs that are more efficient. However, to the best of the authors' knowledge, no experimental evidence of such results has been published. This paper presents experimental results for nine test specimens; five of them were designed using optimal models derived from a full homogenization structural optimization algorithm, and the remaining four using conventional ST models for comparison purposes. Although all specimens carry loads higher than the factored design load, specimens based on ST models derived using full homogenization with reinforcement parallel to the ties exhibit better structural performance regarding crack growth control, more ductile modes of failure and a greater increase in load capacity.