This study focused on two areas under the overarching theme of the effects of domain-general and domain-specific scaffolds with different levels of support, continuous or faded. First, the study investigated the effects of scaffolds on learning of scientific content and problem-solving outcomes. Second, the study examined whether students' prior knowledge and metacognitive skills predict their success in problem solving across different scaffolding conditions. A total of nineteen classes were randomly assigned to one of the four scaffolding conditions: domain-general continuous (DG-C), domain-general faded (DG-F), domainspecific continuous (DS-C), and domain-specific faded (DS-F). Each class had access to different worksheets depending on the scaffolding condition they had been assigned. All students engaged in four problem-solving activities for thirteen class periods. Students' scores on a multiple-choice pretest, posttest, inventory of metacognitive self-regulation, and four recommendation forms were analyzed. iv Results of the study revealed that students' content knowledge in all conditions significantly increased over the thirteen class periods. However, the continuous domainspecific condition outperformed the other conditions on the posttest. Although domaingeneral scaffolds were not as effective as domain-specific scaffolds on learning of scientific content and problem representation, they helped students develop solutions, make strong justifications, and monitor their learning. Unlike domain-specific scaffolds, domain-general scaffolds helped students transfer problem-solving skills even when they were faded. In terms of individual differences, results indicated that while students with lower prior knowledge and lower metacognitive skills benefited from the domaingeneral continuous condition, students with lower regulation of cognition benefited from the domain-general faded condition. Moreover, while students with lower prior knowledge, lower knowledge of cognition, and lower problem representation benefited from the domain-specific continuous condition, students with lower problem representation benefited from the domain-specific faded condition. In contrast, results of the study suggested that scaffolds did not substantially benefit the students with higher prior knowledge and higher metacognitive skills. Several suggestions are discussed for making further improvements in the design of scaffolds in order to facilitate illstructured problem solving in hypermedia learning environments.