Language computational processing is not a straightforward task, most of the Natural Language Processing (NLP) tasks focused on computing underlying sentiments, while emotions are vital components of any language and known to be difficult to detect. Several studies have been carried out in English, but research in Arabic emotion detection is still in its infancy. With the rise of web 2.0 and social media platforms, the amount of textual data with embedded emotions has significantly increased. Although detecting emotions from a text is not trivial, researchers are interested to utilise established artificial intelligence techniques to build highperformance models for this task. This study has been undertaken to provide a Systematic Literature Review (SLR), which is defined as the process of identifying, assessing, and interpreting available resources related to a certain topic to answer the SLR research questions. The aim of this study is to answer questions about textbased Arabic emotion detection challenges and effective methods. Results show that the prevailing challenge in Arabic emotion detection is the limited availability of Arabic emotions annotated training dataset and the morphological complexity and dialect diversity in Arabic. Also, it has been found that most recent studies utilise deep learning approaches.