Background and Aim The effects of very early aphasia therapy on recovery are equivocal. This article examines predictors of very early aphasia recovery through statistical modeling. Methods This study involved a secondary analysis of merged data from two randomized, single-blind trials conducted in Australian acute and subacute hospitals. Study 1 (n = 59) compared daily therapy to usual ward care for up to 4 weeks poststroke in patients with moderate to severe aphasia. Study 2 (n = 20) compared daily group therapy to daily individual therapy for 20 1-hour sessions over 5 weeks, in patients with mild to severe aphasia. The primary outcome measure was the Western Aphasia Battery Aphasia Quotient (AQ) at therapy completion. This analysis used regression modeling to examine the effects of age, baseline AQ and baseline modified Rankin Scale (mRS), average therapy amount, therapy intensity, and number of therapy sessions on aphasia recovery. Results Baseline AQ (p = 0.047), average therapy amount (p = 0.030), and baseline mRS (p = 0.043) were significant predictors in the final regression model, which explained 30% (p <0.001) of variance in aphasia recovery. Conclusion The amount of very early aphasia therapy could significantly affect communication outcomes at 4 to 5 weeks poststroke. Further studies should include amount of therapy provided to enhance reliability of prognostic modeling in aphasia recovery. © 2013 by Thieme Medical Publishers, Inc.