Limited data exist on the burden of influenza in developing countries. In 2008, in order to better understand the epidemiology of influenza virus infection in Tanzania, the Tanzanian Ministry of Health and Social Welfare created a sentinel surveillance system for influenza. At 5 hospitals across the country, patients with influenza-like illness (ILI) and severe acute respiratory illness (SARI) had oropharyngeal and nasopharyngeal samples collected. At the National Influenza Center in Dar es Salaam, specimens were tested for influenza using real-time polymerase chain reaction tests. From May 2008 through November 2010, a total of 1794 samples were collected from 5 sentinel sites, of which 61% were from patients with ILI and 39% were from patients with SARI. Of all ILI and SARI samples, 8.0% were positive for influenza; 6.9% yielded influenza A virus, and 1.1% yielded influenza B virus. Most influenza A virus was subtype H3, which circulated in nearly every month of 2010. The proportion of influenza-positive cases was similar among ILI (8.5%) and SARI (7.3%) patients (P = .39). In multivariate logistic regression, influenza-positive SARI cases were more likely than influenza-negative SARI cases to have had rhonchi (adjusted OR [aOR], 2.31; 95% confidence interval [CI], 1.14-4.67), nasal discharge (aOR, 4.57; 95% CI, 1.30-16.10), and stridor (aOR, 2.63; 95% CI, 1.17-5.90). Influenza-positive ILI patients had a longer duration of fever on presentation, compared with influenza-negative ILI patients (median, 4 vs 3 days; P = .004). Otherwise, there was no difference in signs or symptoms among influenza-positive and influenza-negative ILI patients. During 2.5 years of surveillance for influenza at 5 geographically disbursed sites in Tanzania, we found that influenza circulated year-round. Surveillance should continue in order to fully understand the seasonality and epidemiology of influenza in Tanzania.