A four-stage model of telephone helping was tested by rating 59 calls to a crisis intervention service. The rating instrument was the Crisis Call Interaction Form, a process measure with 19 behaviors in four categories: establishment of a helpful climate, assessment of the crisis, affect integration, and problem solution. Hypotheses stated that helper behaviors would vary differentially across portions (thirds) of calls. After controlling for length of call and within-call similarities, residual variances in each third of the calls were subjected to planned orthogonal comparisons. Hypotheses were generally supported: Climate decreased from first to middle third; assessment decreased in the last third; affect tended to be highest in the middle third; and problem solution increased steadily throughout the calls. However, helper behaviors from all categories were present in every portion of calls. The term "phase" was offered to replace the stage concept as a more fluid model of telephone crisis intervention.