16S rRNA gene profiling (amplicon sequencing) has become a popular technique for understanding host-associated and environmental microbial communities. Illumina sequencing platforms, including the HiSeq and MiSeq, have become the gold standard in amplicon sequencing, and research groups both large and small utilise these tools. The key differences between these platforms pertain to sequencing depth and read length, and while the HiSeq platform has a higher overall cost, large sample sets can take advantage of the high throughput performance to lower the cost per sample. However, the HiSeq platform can prove expensive for research groups analysing small sample sizes, who instead often opt for the Illumina MiSeq platform. With similar chemistry used between the two platforms, it has been widely accepted that sequence output from HiSeq and MiSeq platforms are equivalent and can be seamlessly combined for downstream analyses. However, discrepancies in results from a large study utilising combined HiSeq and MiSeq 16S rRNA data led us to ask again whether output from these two platforms can be combined with confidence for standard microbiome analysis. Using technical replicates of reef-building coral samples from two species, Montipora aequituberculata and Porites lobata, we examined the congruency of these two platforms (HiSeq and MiSeq) with regard to standard alpha and beta diversity metrics to establish comparability of combined datasets. We found some variation in both alpha and beta diversity between sequencing platforms, with inconsistencies according to host species and diversity/dissimilarity metrics that took into account presence/absence, abundance and phylogeny. These inconsistencies in bacterial communities between sequencing platforms persisted following truncation of the dataset to remove low abundance taxa, and also when comparing across higher taxonomic levels. Significant differences in diversity that varied according to dissimilarity metric and host species suggests that bacterial community differences are likely to be context dependent and difficult to correct for without extensive validation work. The results of this study encourage caution in the statistical comparison and interpretation of studies that combine HiSeq and MiSeq rRNA sequence data and point to a need for further work identifying mechanistic differences.