Cucumis melo is highly diverse for fruit traits providing wide breeding and genetic research opportunities, including genome-wide association (GWA) analysis. We used a collection of 177 accessions representing the two C. melo subspecies and 11 horticultural groups for detailed characterization of fruit traits variation and evaluation of the potential of GWA for trait mapping in melon. Through genotyping-by-sequencing, 23,931 informative SNPs were selected for genome-wide analyses. We found that linkage-disequilibrium decays at ~100 Kb in this collection and that population structure effect on association results varies between traits. We mapped several monogenic traits to narrow intervals overlapping with known causative genes, demonstrating the potential of diverse collections and GWA for mapping Mendelian traits to a candidate-gene level in melon. We further report on mapping of fruit shape quantitative trait loci (QTLs) and comparison with multiple previous QTL studies. Expansion of sample size and a more balanced representation of taxonomic groups might improve efficiency for simple traits dissection. But, as in other plant species, integrated linkage-association multi-allelic approaches are likely to produce better combination of statistical power, diversity capture and mapping resolution in melon. Our data can be utilized for selection of the most appropriate accessions for such approaches.