This study aimed to measure accuracy of speech (monosyllables and sentences) and environmental sound recognition in the children with hearing aids (HA, N = 17) or cochlear implants (CI, N = 22) when listening in the classroom noise. Nine normal-hearing children participated as a control group. The classroom noise was recorded either just right before the class or during the break time. All the target sounds were presented at 65 dB SPL, and 5 dB signal-to-noise ratio was used between the target and noise sounds. Additionally, the usage of sign launage, CHAPPS (Children’s Auditory Processing Performance Scale), forward/backward digit span, KISE-KIT (Korea Institute for Special Education-Korea Intelligence Test for children), REVT (Receptive and Expressive Vocabulary Test) results were obtained for the purpose of correlational analysis. Three main findings were observed as follows. First, the type of noise significantly affected recognition accuracy of both HA and CI groups. Closed-set environmental sounds were recognized more accurately than sentence or words for both groups. Overall, children who better recognized the environmental sounds were also superior to recognize speech in noise. Second, children with more usage of sign language as well as worse CHAPPS and digit span scores had lower speech recognition performance in noise, regardless the type of stimulus. Third, the environmental sounds having more repetition pattern and longer duration were more easily recognized for both groups. The analysis of word recognition errors showed that, regardless of group and noise type, the common error types were no response, response as different words, and consonant substitution.