Ethiopia is the motherland and center of genetic diversity of Arabica coffee (<i>Coffea arabica</i> L., Rubiaceae). Receiving extra information on genetic variability is a precondition for more enhancement of coffee (<i>Coffea arabica</i> L.). The present study was carried out to study the correlation and path coefficient analysis of 104 entries consisting of 100 accessions from southern parts of Ethiopia and four standard cultivars were evaluated using augmented design. Data on 22 quantitative traits were recorded from five envoy trees per row for each accession. The result revealed that, for nearly all of the traits phenotypic correlation coefficients were lower than genotypic correlation coefficients, indicative of the inherent union among various traits is less influenced by environment. The genotypic path coefficient analysis revealed that leaf area, number of primary branches, number of nodes of primary branches, canopy diameter, bean width, stem diameter, fruit width, fruit thickness, hundred bean weight, bean thickness and average inter node length of primary branches, had positive direct effects on yield per tree, at the same time as the other traits affected yield indirectly, primarily through total plant height, number of secondary branches, height up to first primary branches, fruit length, bean length, average length of primary branches, leaf width and leaf length, show negative direct sound effects on yield per tree.