Background: Exposure to some phthalate diesters has been associated with adverse reproductive health outcomes in both rodents and humans indicative of anti-androgenic effects. Exposure during sensitive periods of development, such as prenatally, is of particular concern. Objectives: We wished to investigate whether phthalate metabolites measured in maternal serum samples from historical birth cohorts can be used to assess prenatal exposure. Further, we aimed to study temporal and geographical trends in phthalate exposure across three different birth cohorts. Methods: We compared phthalate metabolite levels in maternal serum samples from an Australian (1989–91) and a Danish (1997–2001) birth cohort with levels in serum and urine samples from a recent Danish birth cohort (2012–14). Samples were analysed for 32 phthalate metabolites from 15 phthalate diesters by isotope-diluted liquid chromatography-tandem mass spectrometry (LC-MS/MS). Correlations between metabolites were tested by Spearman rank correlation test, and differences between the cohorts were tested by Mann-Whitney U test. Results: Overall, we observed large variations in serum phthalate metabolite levels between individuals. Secondary metabolites of di-(2-ethyl-hexyl) phthalate (DEHP) and di-iso-nonyl phthalate (DiNP) in serum were weakly to moderately and positively correlated to the levels measured in urine, and secondary metabolites of DEHP were also moderately to strongly and significantly correlated in serum. Correlations with mono-(2-ethyl-hexyl) phthalate (MEHP) and mono-iso-nonyl phthalate (MiNP), the two primary metabolites of DEHP and DiNP, were inconsistent, and we found indications of sample contamination. We observed some significant differences in phthalate metabolite levels between the three cohorts with generally higher levels in the older birth cohorts. Conclusion: Based on comparison across two older birth cohorts and a recent cohort, our results support the concept that historical biobanked serum samples may be used for assessment of prenatal exposure to phthalates when using serum levels of the monoesters of the low-molecular weight (LMW) phthalates and the secondary metabolites of the high-molecular weight (HMW) phthalates. Serum phthalate measurements are, however, not suitable for human biomonitoring and should only be used to exploit historical samples from cohorts, where urine samples were not collected. Our findings suggest that phthalate exposure may have decreased over time from the early 1990s to the 2010s.