Having co-authored a review of the epidemiologic research*, I believe long term (approx. 10 or more years) cell phone use increases tumor risk in humans, especially gliomas, acoustic neuromas, and tumors of the parotid gland.

I agree with mbq that the effect is likely due to electromagnetic radiation (EMR) that interferes with bioelectrical processes through non-thermal mechanisms. 

A cursory review (unpublished) of 116 published toxicology studies, including animal studies and human cellular studies, suggests that some cell phone carrier systems are more likely to be biologically reactive than others. Namely, systems that pulse the signal at extremely low frequencies (i.e., TDMA, GSM, and UMTS [e.g., AT&T and T-Mobile]) are about three times more likely to yield statistically significant biologically reactive outcomes as compared to carriers that do not pulse the signal (i.e., CDMA and W-CDMA [e.g., Verizon and Sprint]) (36% vs. 12%).  However, when EMR is tested at higher power outputs that exceed the current legal limits for cell phones (based upon a measure called the Specific Absorption Rate), then toxic effects are more likely observed regardless of carrier. However, at these higher power outputs we cannot rule out a thermal mechanism.

* Seung-Kwon Myung, Woong Ju, Diana D. McDonnell, Yeon Ji Lee, Gene Kazinets, Chih-Tao Cheng, and Joel M. Moskowitz. Mobile Phone Use and Risk of Tumors: A Meta-Analysis. Journal of Clinical Oncology. 20(27):5565-5572. Published online first Oct 13, 2009. Nov 20, 2009.URL: http://www.ncbi.nlm.nih.gov/pubmed/19826127

Joel M. Moskowitz, Ph.D., UC Berkeley