Research has suggested that long spacing (i.e., temporal intervals) within a training session facilitates second language vocabulary learning. Studies, however, have been limited to treatment that involved sessions for only initial learning but not subsequent relearning. Furthermore, most studies have investigated only the benefits of spacing without considering its potential costs (i.e., increased duration of the treatment). In our study, we examined the benefits and costs of within-session spacing for both initial learning and relearning. In this study, 170 Japanese-speaking university students learned 20 English-Japanese word pairs using one of the following four combinations of initial and relearning spacing: long-long, long-short, short-long, and short-short spacing. The results showed that introducing long spacing for both initial learning and relearning (long-long) led to better long-term retention and higher efficiency scores (i.e., number of words learned per trial) despite the increased duration of the treatment. These findings suggest that the benefits of long spacing outweigh its costs.