Conifers have been the dominant tree species of many forests for more than 200 million years and are currently ecologically and economically extremely important species. Despite their importance there has, to date, been no gymnosperm genome sequence available. We produced a draft assembly of the 20 Gbp Norway spruce (Picea abies) genome. Ab initio gene prediction identified 28,354 well-supported genes representing a gene number similar to the >100 times smaller genome of Arabidopsis thaliana. Analysis of synonymous substitutions per synonymous site (Ks) identified no evidence of a recent whole-genome duplication suggesting that genome expansion resulted from other mechanisms. Repeat analysis showed that the large genome resulted from the slow and steady accumulation of a diverse set of LTR TEs that were not subsequently removed by unequal recombination, as evidenced by a high abundance of complete LTRs with few solo LTRs being identified. We performed low coverage sequencing of Pinus sylvestris, Abies sibirica, Juniperus communis, Taxus baccata and Gnetum gnemon to enable comparative analyses, revealing that the TE diversity is shared among extant conifers. Profiling of 24nt sRNAs, which are known to silence TEs via methylation, was highly tissue-specific and much lower than in other plants. We further identified numerous long (>10,000 bp) introns that arose due to TE insertions and that seem to be shared across gymnosperm species in addition to the genome containing numerous gene-like fragments, most likely representing pseudogenes. We additionally identified 13,031 spruce-specific and 9,686 conserved long non-coding RNAs, 2,719 miRNA candidates and show that the 21nt sRNA population is highly diverse, as reported previously for other conifer species. The availability of a conifer genome will enable further advances in conifer forestry and breeding as well as enabling evolutionary analysis including this previously missing group of land plants.