Abstract: The present study examined effects of reflection, probes, predicate matching, and casual conversation on perception of counselor warmth, threateningness, helpfulness, and quality of counseling relationship. Each of 128 subjects heard one of eight recordings of seven-minute counseling session segments. Subjects then completed a counselor evaluation inventory consisting of 35 statements, each to be rated on a seven-point scale. Hypotheses tested were as follows: (a) Counseling techniques would affect counselor-warmth ratings; (b) counseling technique would affect counselor-threat ratings; (c) counseling technique would affect counselor-helpfulness ratings; (d) counseling technique would affect ratings on two versions of the Counselor Relationship Inventory; (e) four items in the Counselor Relationship Inventory would be answered differentially depending on counseling technique; and, (f) counseling technique would have a differential effect on the total score on the original Counselor Relationship Inventory II, indicating instrument bias. Effects of counseling technique on the dependent variable scales were tested using seven one-way analyses of variance with Scheffe multiple ranges tests. Counseling technique was found to differentially affect perception of warmth, threat, helpfulness, and both relationship scales. Predicate-matching received higher warmth ratings than reflection or non-counseling, and was rated less threatening than reflections and probes. Predicate-matching was also rated most helpful of the four techniques. Non-counseling conversation was rated least helpful. Predicate-matching also received higher ratings on the relationship scales than reflections or probes. Comparison of scores on a four-item subscale of the Counselor Relationship Inventory with scores on four items designed to eliminate pro-reflection bias indicated that the original items were answered more favorably for reflective counselors than for predicate- matchers. However, total score on the Counselor Relationship Inventory was not significantly affected. It was concluded that item-bias was not of sufficient magnitude to effect instrument-bias. Correlation coefficients indicated that the short scales for warmth, threat, and helpfulness were internally consistent. However, two items in the original Counselor Relationship Inventory were found to be non- significantly correlated with total inventory score.