Clothes affect everyone; we wear them for all occasions; they silently communicate on our behalf, and they can enhance our level of confidence and comfort. For the garment to be comfortable, the garment fit has to be appropriate. Appropriate for the intended function and in line with the wearer’s preferences. For the garment to end up as an approved garment in the customer’s wardrobe, it has to be improved and evaluated many times over. The final evaluation to pass is when the customer finds the garment, tries it on and asks: Does it really fit? The common denominator for the studies included in this thesis is garment fit; the goal is to investigate some methods to improve, find and evaluate garment fit. To improve garment fit, two studies were done. One study is on improving garment fit with the help of a systematic model, based on anthropometric and garment numerical data; this is explored with the help of an experimental set up. The second study is on improving garment fit for the unique figure by offering made-to-measure garments, which is investigated with a structure of action research. An experimental strategy is used to find garments that fit, where the size and fit correspondence is compared between virtual and real garments. To tie everything together, variables for garment fit evaluation are identified with the help of a structured literature review and then analysed within each study. The result shows that the theoretical garment fit improves by using the systematic model. Both the overall accommodation for the target group increased as well as the fit value. The garment fit is improved for the unique figure; this is achieved through complex body measurements, invasive pattern modifications and garment make-up for fit evaluation. The accuracy for size selection based on virtual garments exceeded the one based on the more traditional key measurements. The variables involved in the garment fit evaluation can be divided into five areas: influencing factors, evaluations focus, resources, evaluators or fitting sessions.