Purpose: This clinical trial tested the null hypothesis that there would be no difference in prosthetic maintenance for two-implant mandibular overdentures retained by either a bar-clip mechanism or ball attachments. Prosthetic outcomes are reported over 3 years using a six-field protocol. Materials and Methods: One hundred edentulous participants received new maxillary complete dentures and a mandibular two-implant overdenture (IOD), with random assignment to either a bar and metal clip or two ball attachments (titanium alloy matrix and spring) for retention. Eighty-seven subjects were available for follow-up after 3 years. Results: Almost three times as many bar-clip dentures (63%) were rated successful compared to the ball attachment design. Two percent of the participants in each group died over the course of the study, while 15% of the bar-clip and 8% of the ball IOD subjects were lost to follow-up. More than three times as many ball attachment IODs (60%) required retreatment in the form of excessive repairs, and twice as many of the ball attachment design (8%) required replacement. The ball attachment IOD was significantly more likely to require patrix tightening or matrix replacement, while the barclip design was more likely to require activation of the matrix. Conclusion: Using the criteria of a six-field protocol for implant overdenture outcomes, the bar-clip IOD was a significantly more successful prosthesis, requiring less maintenance than the titanium alloy matrix and spring ball attachment IOD employed in this study. The null hypothesis was therefore defeated. Int J Prosthodont 2003;16:255–260.