The phylogenetic relationships among the ca. 140 species of goldenrods (Solidago; Asteraceae) have been difficult to infer due to species richness, polyploidy, and shallow genetic divergence among species.  We addressed these obstacles by combining aggressive sampling of diploid goldenrod herbarium specimens with the use of a custom Solidago hybrid-sequence capture probe set.  A set of tissues from herbarium samples comprising ca. 90% of Solidago species was assembled and DNA was extracted.  A custom hybrid-sequence capture probe set was designed using two Solidago species, and data from 854 nuclear regions was obtained and analyzed from 209 specimens.  Maximum likelihood and coalescent approaches were used to estimate the species tree for 157 diploid samples.  Hybrid enrichment and sequencing from goldenrod herbarium specimens was successful.  Although DNAs from older specimens were both more fragmented and produced fewer sequencing reads, there was no relationship between specimen age and our ability to obtain sufficient data at the target loci.  The resulting Solidago species tree was generally well supported, with 88/155 (57%) nodes receiving ≥95% bootstrap support.  Solidago was strongly supported as monophyletic, with Chrysoma pauciflosculosa identified as sister.  A clade comprising Solidago ericameriodes, Solidago odora, and Solidago chapmanii was identified as the earliest diverging Solidago lineage.  The previously segregated genera Brintonia and Oligoneuron were identified as placed well within Solidago.  These and other phylogenetic results were used to establish four subgenera and fifteen sections within the genus.  The combination of expansive herbarium sampling and hybrid-sequence capture data allowed us to quickly and rigorously establish the evolutionary relationships within this difficult, species rich group.  This general strategy should allow the phylogenies of similar genera to be inferred.