diff --git a/sample_data/README.md b/sample_data/README.md
new file mode 100644
index 00000000..b39aa7cd
--- /dev/null
+++ b/sample_data/README.md
@@ -0,0 +1,26 @@
+# Unicycler sample data
+
+This sample dataset is made of synthetic reads generated from three plasmids in a [_Shigella sonnei reference_](https://www.ncbi.nlm.nih.gov/genome/417?genome_assembly_id=166795).
+
+
+### Short read-only assembly
+
+`unicycler -1 short_reads_1.fastq.gz -2 short_reads_2.fastq.gz --no_long -o output_dir`
+
+This command will assemble the short reads alone. If you look at the resulting contigs (or view the graph in [Bandage](https://github.com/rrwick/Bandage)) you'll see that only the smallest plasmid assembled completely. The larger two plasmids contain quite a lot of repetitive sequence and short reads aren't enough to complete them.
+
+
+### Low-depth long read hybrid assembly
+
+`unicycler -1 short_reads_1.fastq.gz -2 short_reads_2.fastq.gz -l long_reads_low_depth.fastq.gz -o output_dir`
+
+This command will assemble the short reads along with a small number of long reads. They have an average depth of approximately 1x, so some parts of the plasmids will be represented in these reads but other parts will not.
+
+These reads are not sufficient to complete the whole assembly, we are getting closer. The second smallest plasmid should now be finished and only the largest plasmid remains incomplete.
+
+
+### High-depth long read hybrid assembly
+
+`unicycler -1 short_reads_1.fastq.gz -2 short_reads_2.fastq.gz -l long_reads_low_depth.fastq.gz -o output_dir`
+
+This command will assemble the short reads along using long reads of approximately 20x depth. In this case, all parts of the plasmids are represented in a number of long reads, so there is definitely sufficient information to complete the assemblies. Accordingly, you should now see that the assembly produces just three contigs: one for each plasmid.