The objectives of this study were to assess phenotypic variability among F3 and BC1F2 tomato populations, and apply genotype by yield*trait (GYT) biplots for population and line selection based on multiple traits. Four diverse cultivated parents (‘CLN2498D’ and ‘CLN2417H’ from Ethiopia; ‘UC Dan INDIA’ and ‘Tima’ from Nigeria), and wild parent ‘LA2093’ were used to generate 276 potential breeding lines. The lines were categorized into eight populations (‘Pop_1_W/H1’, ‘Pop_2_W/H2’, ‘Pop_3_W/D1’, ‘Pop_4_W/D2’, ‘Pop_5_W/T1’, ‘Pop_6_W/T2’, ‘Pop_7_W/U1’, and ‘Pop_8_W/U2’), and evaluated twice in the field using 19 × 15 alpha-lattice design with two replicates. Significant differences were observed among lines and populations for all yield enhancing traits. ‘Pop_1_W/H1’, ‘pop_4_W/D2’ and ‘pop_6_W/T2’ expressed the highest genetic divergence for plant height, number of leaves, total flower and fruit number, and fruit weight. GYT biplots revealed that all yield*trait interactions had a positive correlation with each other. F3 populations, ‘Pop_5_W/T1’ and ‘pop_1_W/H1’ exhibited the best performance for majority of the yield*trait combinations. Hierarchical clustering on principal components (HCPC) revealed overlapping lines (70.58% of Cluster D lines) and (54.05% of Cluster U lines) from the two F3 populations. In BC1F2 population, 32.35% of the 34 original lines of Cluster D and 48.48% of Cluster T lines overlapped between Clusters D and T, while 18.18% of Cluster T lines and 8.82% of Cluster H lines were transgressive between Clusters T and H. Transgressive segregants ‘0210U1’, ‘0211U1’, and ‘0171T1’ of selfed population using multivariate analysis were believed to represent potential sources of novel genetic variation for future tomato breeding.