Most surveillance programs for avian influenza (AI) virus in wild birds utilize molecular tests such as real-time reverse transcription-PCR (RRT-PCR) or virus isolation (VI) in embryonating chicken eggs. To provide insight into the relationship between positive diagnostic test results and infectivity for an avian host, we challenged Mallards (Anas platyrhynchos) with Mallard-derived cloacal swab field samples found positive by VI or RRT-PCR. Six of 11 samples that were both RRT-PCR positive and VI positive infected Mallards. Sample infectivity for Mallards appeared to be dependent on concentration of infectious virus in the sample; five of the six samples that replicated in Mallards had a measurable virus titer, whereas four of the five samples that did not infect Mallards had titers below the limit of detection (10(0.9) median embryo infectious dose/0.2 mL). None of seven samples that were RRT-PCR positive and VI negative infected Mallards. These results indicate that embryonating chicken eggs are a sensitive diagnostic tool for detecting Mallards excreting infectious AI virus at a high enough concentration to infect another Mallard; however, not all cloacal swab field samples that are positive by VI or RRT-PCR are infective to another Mallard. Additionally, our results indicate that Mallards are susceptible to Mallard-origin AI viruses that have not been propagated in embryonating chicken eggs and that some of these virus strains can infect birds at titers that are lower than those typically used in experimental challenge studies. These data highlight a need to examine the effects of using egg-propagated AI viruses in experimental trials.