Abstract Exploiting yield heterosis under low soil phosphorus (P) inputs is a possible strategy for enhancing rice productivity and resource use efficiency in the changing climate scenario. In the present study, Line (L) × Tester (T) analysis was conducted with five restorers, three maintainer lines and 15 derived hybrids. The hybrids and parents were evaluated for two wet seasons ( Kharif - 2017 and 2019) across the graded levels of applied P (low—P20 kg ha −1 ; Moderate—P40 kg ha −1 ; Normal—P60 kgha −1 ). Pooled analysis across years revealed highly significant variances in the interaction of the General Combining Ability (GCA) of parents and the Specific Combining Ability (SCA) of hybrids with a predominance of non-additive genetic variance. Testers contributed higher variation for the key traits under graded P conditions. Among the parental lines, CRMS32B showed strong GCA effects for single plant yield (SPY) and number of productive tillers per plant (NPTP), while IR79156B was promising for spikelet fertility percentage (SFP). Among testers, ATR305 and TCP795 exhibited significant GCA effects for SPY, SFP and NPTP and early flowering. The Hybrids, H14 (CRMS32A × ATR226) showed the highest SCA effects for NPTP, SPY and SFP, while H10 (IR79156A × TCP795) exhibited significant SCA effect for earliness. Compared to standard check variety Kasalath, the hybrids demonstrated a 120 to 200% heterosis gain across the graded P levels for the key traits. AMMI analysis identified hybrids, H3 (APMS6A × ATR305), H13 (CRMS32A × ATR305) and H11 (CRMS32A × AYT21) as highly stable performers across graded P levels, while H14 (CRMS32A × ATR226) with moderate stability. The present findings highlight the breeding rice hybrids suited for low P conditions to improve phosphorus use efficiency and ensure yield stability under resource-constrained environments.