This study tested the 'loss spiral' hypothesis of work-home interference (WHI). Accordingly, work pressure was expected to lead to WHI and exhaustion, and, vice versa, exhaustion was expected to result in more WHI and work pressure over time. Results of SEM-analyses using three waves of data obtained from 335 employees of an employment agency offered strong support for this hypothesis. It was found that T1 work pressure and exhaustion were determinants of T2 and T3 WHI, whereas T1 WHI was a causal determinant of T2 and T3 exhaustion and work pressure. In addition, work pressure and exhaustion had causal and reversed causal relationships over time. These empirical findings suggest that common theoretical models postulating the causal chain of work pressure ¿ WHI ¿ exhaustion are inadequate. Rather, more elaborated models including reciprocal relationships between work characteristics, WHI and employee well-being seem more appropriate.