ABSTRACT: The Œgrowth-mortality¹ hypothesis, which holds that larger and/or faster growing individuals will have a higher probability of survival, currently includes 3 functional mechanisms (hypotheses) in its theoretical framework: Œbigger is better¹,Œstage duration¹ and the recently proposed Œgrowth-selective predation¹, which are based on size, time and per se growth rate, respectively. Through otolith microstructure analysis, we tested these 3 synergistic growth-related mechanisms according togrowth characteristics of the survivors vs the original population in the short-term (ca. 2 wk) survival process of larval Japanese anchovy Engraulis japonicus in the Œshirasu¹ (larval anchovy) fishing ground in Sagami Bay, Japan. Back-calculatedstandard length (growth trajectory) and growth rate (growth history) were compared between the survivors (SV) captured on 18 July 2001 and the presumed original population (OP) captured on 1 and 5 July 2001. The larvae from SV were consistently smallerthan the larvae from OP until at least the start of the ca. 2 wk survival process (1 July). Daily growth rates, however, were higher for SV than for OP at least at the start of the survival period. Therefore, faster growing individuals survived even ifthey were smaller than slower-growing conspecifics. This was probably mediated by predation. Growth histories were generally similar between the metamorphosing larvae and non-metamorphosing larvae older than 40 d, the minimum age for metamorphosis, exceptfor the period immediately after hatching. As such, we failed to detect a clear relationship between growth rates and the timing of metamorphosis (stage duration) as a whole. The results supported and extended the Œgrowth-selective predation¹ hypothesisbut not the Œbigger is better¹ hypothesis. The Œstage duration¹ hypothesis was not unequivocally supported by the present findings.