Progress in coarse rice improvement for Pakistan's northern belt has been slowed by the narrow genetic base of adapted materials and by a habit of evaluating candidate lines one trait at a time. The present study takes a multivariate view: eight advanced coarse rice lines (Line-11, Line-20, Line-23, Line-24, Line-25, Line-26, Line-27 and Line-28) were grown in a randomized complete block design with three replications at the Agricultural Research Station, Baffa, during the 2024 kharif season, and eleven agronomic and yield attributes were recorded. Analysis of variance detected highly significant differences among genotypes for every trait (p < 0.01). Broad-sense heritability was high (≥ 0.80) for ten of the eleven traits, and genetic advance as a percentage of the mean exceeded 25% for plant height, flag leaf area, panicle length, tillers per plant, primary branches, secondary branches and 1000-grain weight indicating that additive gene action predominates and that direct phenotypic selection would be effective. Pearson correlation revealed that grain yield was most strongly associated with stem diameter (r = 0.72) and 1000-grain weight (r = 0.55), while tiller number carried a negative correlation (r = –0.61), consistent with the sink-limited nature of coarse rice under the site's cool nights. The first two principal components together explained 62.3% of the total variation; PC1 separated architecture from earliness, while PC2 separated grain-weight-driven yield from vegetative bulk. Ward's hierarchical clustering grouped the eight lines into three well-defined clusters. A weighted selection index that combined yield, grain weight, tiller number, branching and (as a penalty) stem-borer incidence ranked Line-23, Line-20 and Line-26 as the three best-adapted candidates for Hazara Division. Line-11 is retained as a donor for earliness and Line-25 as a donor for secondary-branch density. The results support a shift toward index-based, multi-trait selection in the on-going coarse rice programme for the region.