Maternal immune activation (MIA) models that are based on administration of the viral mimetic, poly(I:C), are widely used as experimental tools to study neuronal and behavioral dysfunctions in relation to immune-mediated neurodevelopmental disorders and mental illnesses. Evidence from investigations in non-pregnant rodents suggests that different poly(I:C) products can vary in terms of their immunogenicity, even if they are obtained from the same vendor. The present study aimed at extending these findings to pregnant mice, while also controlling various poly(I:C) products for potential contamination with lipopolysaccharide (LPS). We found significant variability between different batches of poly(I:C) potassium salt obtained from the same vendor (Sigma-Aldrich) in terms of the relative amount of dsRNA fragments in the high molecular weight range (1000–6000nucleotides long) and with regards to their effects on maternal thermoregulation and immune responses in maternal plasma, placenta and fetal brain. Batches of poly(I:C) potassium salt containing larger amounts of high molecular weight fragments induced more extensive effects on thermoregulation and immune responses compared to batches with minimal amounts of high molecular weight fragments. Consistent with these findings, poly(I:C) enriched for high molecular weight dsRNA (HMW) caused larger maternal and placental immune responses compared to low molecular weight (LMW) poly(I:C). These variable effects were unrelated to possible LPS contamination. Finally, we found marked variability between different batches of the poly(I:C) potassium salt in terms of their effects on spontaneous abortion rates. This batch-to-batch variability was confirmed by three independent research groups using distinct poly(I:C) administration protocols in mice. Taken together, the present data confirm that different poly(I:C) products can induce varying immune responses and can differentially affect maternal physiology and pregnancy outcomes. It is therefore pivotal that researchers working with poly(I:C)-based MIA models ascertain and consider the precise molecular composition and immunogenicity of the product in use. We recommend the establishment of reference databases that combine phenotype data with empirically acquired quality information, which can aid the design, implementation and interpretation of poly(I:C)-based MIA models.