Work passion is related to adaptive outcomes in the work context (Moeller et al., 2019; Pollack et al., 2020), therefore its preservation (Hardgrove, 2019) or it’s enhancement could be desirable. Yet the mechanisms involved in passion fostering are only understood insufficiently (Cameron, 2019; Chummar, Moeller, 2021; Singh, & Ezzedeen, 2019; Pury, & Hardy, 2016), particularly in regard to appropriate timeframes and antecedents (Cameron, 2019). Thus, a four wave longitudinal study (1 week spacing; N=338) from April to August 2020 in Germany investigated relationships of basic need satisfaction (BNS; autonomy, competency, relatedness; Deci & Ryan, 2000) in two domains (work & leisure) with harmonious (HWP) and obsessive (OWP) work passion (Dualistic model of passion, DMP, Vallerand et al, 2003; 2019) over time. Multilevel longitudinal modelling of the German working student sample with autoregressive effects showed a relationship for autonomy (work) and HWP over time, and a negative relationship of autonomy (leisure) and OWP over time, only. Cross-sectional results using all measurement occasions are mostly in line with other studies, showing relationships of (work) autonomy, competency and relatedness with HWP, as well as a relationship of (work) relatedness with OWP and a negative relationship of (leisure) relatedness with OWP. Some relationships, however, fail the significance threshold when applying Bonferroni correction. Results are only partially in line with DMP theory, yet mostly in line with previous findings (e.g. Curran et al., 2015 ). Limitations are the focus on that particular timeframe and a predominantly white-collar student worker sample. Theoretical implications include the mixed empirical evidence in relation to predictions made by DMP, practical implications include the usability of results in interventions studies. Contributions are the evaluation of relevant passion malleability timeframes and the differentiated investigation of BNS in preparation for intervention studies, that many other studies are lacking. The 'ready for analysis' data provided here is in the LONG format, devoid of demographic variables, and available as a .CSV file, as well as a SPSS data file. The latter was used for the analysis of the according publication with the R syntax also provided in this repository. So you can either run the SPSS data file with the R syntax "as is", or adapt the R syntax to import the .CSV file to replicate the analysis. Furthermore, the unprepared and unfiltered data (devoid of demographic variables and free form text input variables) is attached as a .CSV file.