Background: Dengue fever (DF) is the most common mosquito borne viral disease, with an estimated 390 million infections a year, globally. In 1994, Saudi Arabia reported their first confirmed case of DF in the western coastal city of Jeddah. Over the next couple of decades, the disease spread to other parts of the country which led to the Saudi Ministry of Health declaring DF endemic in the western region of Saudi Arabia. Today, Saudi Arabia has one of the largest DF burdens in the Middle East, with the most recently published incidence rate 16 per 100,000 people. While a breadth of scientific literature has demonstrated strong associations between weather and DF, there is very little research into the topic in this region of the world. Saudi Arabia’s extreme desert climate, as well as its unique position as host country of the world’s largest annual mass gathering event, the Hajj, presents a unique opportunity to explore the relationships between weather, large population movements between areas with high DF prevalence, and DF in a previously unstudied region. Infectious disease modelling is commonly used to examine and quantify associations between DF and influencing factors, and predict disease incidence, supporting public health systems in DF prevention efforts by providing insight into disease ecology and early warning of increased disease activity. Methods: We first performed a systematic review of the literature on DF in Saudi Arabia, and the environmental and population factors that may have played a role in DF emergence and subsequent spread and endemicity. Based on the review, we selected a number of weekly weather variables, namely temperature, humidity and rainfall, and population variables that describe the annual Hajj pilgrimage, to statistically explore their association with weekly DF incidence records. We created and tested statistical predictive models using three different approaches; poisson multivariate regression, ARIMA, and random forest regression, and compared their performance using R2 and RMSE as measures of correlation and error, respectively. We also applied these local data points to a previously developed and validated process based dynamic model (DyMSiM). We further built on this model by incorporating processes that describe population movement during the annual Hajj pilgrimage (DyMSiM(P)). The dynamic models were also evaluated using the same measures of correlation and error. Results: The systematic review identified temperature, humidity, and rainfall as the environmental factors most likely to influence DF incidence, and supported the hypothesis of the Hajj pilgrimage as a potential source of virus importation. A bivariate analysis revealed temperature and relative humidity to have the strongest correlations with DF incidence. Among the three predictive models tested, the random forest model performed the best. The DyMSiM model performance was variable, but incorporation of pilgrimage data points using DyMSiM(P) improved overall performance. Discussion: The results of the bivariate analyses between DF and weather variables support previous work locally and globally that suggest that temperature and humidity play an important role on disease incidence, likely due to their effect on mosquito population dynamics. The influence of precipitation was less accentuated most probably because the primary breeding habitat for mosquitoes in urban areas such as Jeddah are indoor containers. The association with pilgrimage variables in the bivariate analyses pointed towards the Hajj as an important source of virus importation and introduction of novel DENV serotypes, but quantifying the association is complicated by biases on disease reporting due to stress on public health resources during this period. The short study period compounded by the seasonal variability of the timing of the annual Hajj season further limited our ability to accurately evaluate this association. Among the three predictive models, the random forest model performed the best primarily due to its ability to capture non-linear associations and has the most practical application as it is not dependent on surveillance data. Both process-based models DyMSiM and DyMSiM(P) performed better during years of more moderate temperature and humidity variables which is explained by the effect of extreme temperature and humidity on mosquito populations. This suggests that in geographical areas such as Saudi Arabia where extremes of weather are common, it is very likely that traditional sources of weather data do not reflect weather variables in the urban pockets where mosquitoes flourish. While the addition of pilgrimage data points to the dynamic model improved performance this effect was blunted by the profound influence of high temperature and humidity on the mosquito population. Conclusion: Temperature and humidity play an important role on DF transmission in Saudi Arabia. Among the predictive models evaluated the model using a random forest approach performed the best at predicting DF in this region. The DyMSiM model applied to this dataset also performed relatively well, particularly when altered to include data points describing the Hajj. These findings suggest that large scale population movement as occurs during the pilgrimage likely leads to virus importation and introduction of new virus strains. There is potential for further refinement of these associations with extension of the study period and with the acquisition of on the ground weather data that more accurately reflects mosquito habitats in large urban areas.