Four treatments were established to evaluate the effect of delayed first exogenous feeding and subsequent feeding periods on the development of goldfish larvae. The fasting and feeding periods were arranged as follows: T1 (0DFA:30DF), T2 (4DFA:26DF), T3 (8DFA:22DF), and T4 (12DFA:18DF), where DFA are the days of fasting and DF are the days of feeding. The larvae were kept in community tanks for each repetition, with 75 larvae per tank (3.75 larvae L-1). There was no significant difference in total length or final weight between larvae submitted to 0, 4, and 8 days of feed deprivation. However, larvae submitted to 12 DFA had higher values of the same variables than the other treatments. No significant difference in larval survival was observed between T1, T2, and T3, but T4 negatively influenced survival. Under the conditions evaluated, the point of no return was 4 DFA. Dead larvae were partly eaten in the fasting treatments. The delayed first feeding should be avoided in large-scale productions since it significantly reduces survival during cultivation. This article results from research funded by the Pescarte Environmental Education Project (PEA/IBAMA).