Breeding of cucumber for protected cultivation requires improvement of multiple complex horticultural traits, but studies combining multi-trait association mapping with genomic prediction in the same breeding panel remain limited. We evaluated 208 cucumber inbred lines independently developed from diverse source materials collected across China. Sixteen agronomic traits covering vegetative growth, reproductive development, fruit characteristics, yield, and disease resistance were measured over two consecutive growing seasons, and whole-genome resequencing uncovered 580,151 high-quality single nucleotide polymorphisms (SNPs). Genome-wide association analysis detected 89 significantly associated loci (SALs) for 15 traits, including 10 genomic regions shared by two or more traits. Beyond the known CsFT and Tu/CsDULL loci, numerous additional SALs were identified, yielding 19 candidate genes for further investigation of the corresponding traits in cucumber. Genomic prediction accuracies using rrBLUP were evaluated for nine quantitative traits and were variable among traits, with fruit length, fruit shape index, and fruit neck length showing relatively high prediction accuracies. Together, these results have furthered our understanding of the genetic architecture and genomic predictability of cucumber agronomic traits and laid a foundation for integrating locus-specific and genome-wide information for future breeding strategies.