Abstract: In order to study the relations among lint yield and yield components in cotton chromosome segment substitution lines (CSSLs) and clarify the main direction of yield breeding of cotton, a set of CSSLs and its F1 have been used to analysis the relations among lint yield and yield components using the correlation and path analysis in five environments. The results showed that the coefficient of variation of boll number among yield components was the largest, and followed by lint percentage and boll weight, respectively. Compared to CSSLs, the coefficient of variation of yield components in F1 was smaller in all environments. Lint yield exhibited positive and significant correlations with boll weight, lint percentage and boll number except for a few environments. The coefficients of simple correlation and partial correlation of boll number in CSSLs were the largest, but the coefficients of lint percentage in F1 were the largest. Compared to lint percentage and boll number, the relation between boll weight and lint yield was controlled significantly by environment. Path analysis illustrated that the boll number in CSSLs had the highest positive direct effect on lint yield, followed by lint percentage and boll weight, but the lint percentage in F1 had the highest positive direct effect on lint yield, followed by boll number and boll weight except for one environment. The indirect effects of yield components on lint yield in CSSLs were positive, but the indirect effects of boll number via boll weight and boll weight via boll number on lint yield in F1 were negative. The results indicated that the relations between lint yield and yield components, and the effects of yield components on lint yield were significantly controlled by the type of materials and environment. Based on the results of this study, the main direction of cotton yield breeding using CSSLs was to improve lint percentage at the early stage and boll number at the later stage, and For F1, the main direction of yield breeding was to improve lint percentage.