diff --git a/html/import.html b/html/import.html
index aee543927..35e9ec83f 100644
--- a/html/import.html
+++ b/html/import.html
@@ -34,6 +34,30 @@
               <br />
               <h2 class="title">Data Import</h2>
               <div class="content">
+                <p>
+                  Tasks can be imported from files with this command:
+
+                  <pre><code>% task import file</code></pre>
+
+                  A variety of different file types are recognized by task, namely:
+
+                  <ul>
+                    <li>Tasks exported from task prior to version 1.5.0.
+                    <li>Tasks exported from task version 1.5.0 and later.  The file
+                        format changed with 1.5.0.
+                    <li>todo.sh files.
+                    <li>CSV files with a variety of recognized column names.
+                    <li>Plain text files, with one task listed per line.
+                    <li>Task command line format.
+                  </ul>
+
+                  Task makes a good effort to determine which of these formats a
+                  file is, and then imports accordingly.
+                </p>
+
+                <p>
+                  It would be wise to backup your task data files before an import.
+                </p>
               </div>
 
               <br />
diff --git a/src/import.cpp b/src/import.cpp
index be3dab21e..0dc2ba4a8 100644
--- a/src/import.cpp
+++ b/src/import.cpp
@@ -240,8 +240,13 @@ std::string handleImport (TDB& tdb, T& task, Config& conf)
     std::vector <std::string> lines;
     slurp (file, lines, true);
 
+    // Take a guess at the file type.
+    fileType type = determineFileType (lines);
+
+    // TODO Allow an override.
+
     // Determine which type it might be, then attempt an import.
-    switch (determineFileType (lines))
+    switch (type)
     {
     case task_1_4_3:    out << importTask_1_4_3  (tdb, conf, lines); break;
     case task_1_5_0:    out << importTask_1_5_0  (tdb, conf, lines); break;