diff --git a/app/Support/HtmlHelper.php b/app/Support/HtmlHelper.php
index d57893a2..249b152e 100644
--- a/app/Support/HtmlHelper.php
+++ b/app/Support/HtmlHelper.php
@@ -80,6 +80,36 @@ public static function sanitizeHtml(?string $html): string
         return self::htmlSanitizer()->sanitize($html);
     }
 
+    /**
+     * Render `libri.descrizione`, which may be either editor HTML or legacy
+     * plain text with newlines (imported records).
+     *
+     * Plain text gets nl2br so its line breaks survive. Content that is already
+     * block-structured HTML is left as-is: the editor (TinyMCE) separates its
+     * <p> paragraphs with a literal "\n", and running nl2br over that injects a
+     * stray <br> between every block, double-spacing the output (issue #313).
+     * The result is always sanitized. This intentionally remains specific to
+     * the book-description field addressed by issue #313.
+     *
+     * @param string|null $value Raw stored value
+     * @return string Sanitized HTML ready to echo
+     */
+    public static function bookDescription(?string $value): string
+    {
+        if ($value === null || $value === '') {
+            return '';
+        }
+
+        // If the value already carries block/line HTML, its line breaks are
+        // markup — nl2br would turn the newlines *between* blocks into <br>.
+        $hasBlockHtml = preg_match(
+            '/<(?:p|br|div|ul|ol|li|h[1-6]|blockquote|pre|table|hr)\b/i',
+            $value
+        ) === 1;
+
+        return self::sanitizeHtml($hasBlockHtml ? $value : nl2br($value, false));
+    }
+
     /**
      * Return a public absolute HTTP(S) URL, or an empty string when unsafe.
      *
diff --git a/app/Views/frontend/book-detail.php b/app/Views/frontend/book-detail.php
index be4800fd..0e3e9227 100644
--- a/app/Views/frontend/book-detail.php
+++ b/app/Views/frontend/book-detail.php
@@ -2026,7 +2026,7 @@ class="book-cover-large img-fluid"
                                         : \App\Support\MediaLabels::formatTracklist($musicDescription) ?>
                                 </div>
                             <?php else: ?>
-                                <div class="prose prose-sm max-w-none"><?= \App\Support\HtmlHelper::sanitizeHtml(nl2br($book['descrizione'], false)) ?></div>
+                                <div class="prose prose-sm max-w-none"><?= \App\Support\HtmlHelper::bookDescription($book['descrizione']) ?></div>
                             <?php endif; ?>
                         <?php else: ?>
                             <p class="text-gray-500"><?= __("Nessuna descrizione disponibile per questo libro.") ?></p>
diff --git a/app/Views/libri/scheda_libro.php b/app/Views/libri/scheda_libro.php
index f9a11671..c06831bb 100644
--- a/app/Views/libri/scheda_libro.php
+++ b/app/Views/libri/scheda_libro.php
@@ -718,7 +718,7 @@ class="inline-flex items-center gap-1 px-2 py-1 text-xs font-medium rounded-full
               ?>
               <?= \App\Support\MediaLabels::formatTracklist($descText) ?>
             <?php else: ?>
-              <?php echo App\Support\HtmlHelper::sanitizeHtml(nl2br($libro['descrizione'], false)); ?>
+              <?php echo App\Support\HtmlHelper::bookDescription($libro['descrizione']); ?>
             <?php endif; ?>
           </div>
         </div>
diff --git a/tests/bookdescription-render-313.unit.php b/tests/bookdescription-render-313.unit.php
new file mode 100644
index 00000000..b8dcdf75
--- /dev/null
+++ b/tests/bookdescription-render-313.unit.php
@@ -0,0 +1,58 @@
+<?php
+declare(strict_types=1);
+
+/**
+ * Reusable guards for the issue #313 fix (OPAC description double-spacing).
+ * Complements richtext-nl2br-313.unit.php with edge cases and a render-site
+ * source guard so the description field can't silently regress to raw nl2br().
+ *
+ * Run:  php tests/bookdescription-render-313.unit.php   (exit 0 iff all pass)
+ */
+
+$root = dirname(__DIR__);
+require $root . '/vendor/autoload.php';
+
+use App\Support\HtmlHelper;
+
+$pass = 0;
+$fail = 0;
+$check = static function (bool $ok, string $label) use (&$pass, &$fail): void {
+    if ($ok) { $pass++; echo "  OK  {$label}\n"; }
+    else     { $fail++; echo "  FAIL {$label}\n"; }
+};
+
+// 1. Windows CRLF plain text keeps one <br> per line break (not two).
+$crlf = "Prima riga\r\nSeconda riga";
+$out = HtmlHelper::bookDescription($crlf);
+$check(substr_count($out, '<br') === 1, "01 CRLF plain text yields exactly one <br> per newline");
+
+// 2. HTML with attributes on the block tag is still detected as HTML (no <br>).
+$attr = "<p class=\"lead\">Uno.</p>\n<p>Due.</p>";
+$out = HtmlHelper::bookDescription($attr);
+$check(!str_contains($out, '<br'), "02 block tag with attributes is treated as HTML (no injected <br>)");
+
+// 3. A leading plain sentence before a block still counts as HTML overall
+//    (has a block marker), so its internal newline is not turned into <br>.
+$mixed = "Intro\n<p>Blocco.</p>";
+$out = HtmlHelper::bookDescription($mixed);
+$check(!str_contains($out, '<br'), "03 value containing any block marker suppresses nl2br");
+
+// 4. Render source guard: the OPAC book detail renders the description through
+//    the helper, never raw nl2br() — the exact regression #313 fixed.
+$bookDetail = (string) file_get_contents($root . '/app/Views/frontend/book-detail.php');
+$check(
+    str_contains($bookDetail, 'HtmlHelper::bookDescription($book[\'descrizione\'])')
+        && !preg_match('/nl2br\(\s*\$book\[\'descrizione\'\]/', $bookDetail),
+    "04 book-detail renders \$book['descrizione'] via bookDescription(), not raw nl2br()"
+);
+
+// 5. Same guard for the admin book sheet.
+$scheda = (string) file_get_contents($root . '/app/Views/libri/scheda_libro.php');
+$check(
+    str_contains($scheda, 'HtmlHelper::bookDescription($libro[\'descrizione\'])')
+        && !preg_match('/nl2br\(\s*\$libro\[\'descrizione\'\]/', $scheda),
+    "05 admin scheda renders \$libro['descrizione'] via bookDescription(), not raw nl2br()"
+);
+
+echo "\n{$pass} PASS, {$fail} FAIL\n";
+exit($fail === 0 ? 0 : 1);
diff --git a/tests/richtext-nl2br-313.unit.php b/tests/richtext-nl2br-313.unit.php
new file mode 100644
index 00000000..db70d9d4
--- /dev/null
+++ b/tests/richtext-nl2br-313.unit.php
@@ -0,0 +1,72 @@
+<?php
+declare(strict_types=1);
+
+/**
+ * Regression test for issue #313 — the OPAC injected <br> between the <p>
+ * paragraphs of a book description.
+ *
+ * Root cause: the book description was rendered with
+ * sanitizeHtml(nl2br($value, false)). Editor HTML separates its <p> blocks
+ * with a literal "\n", so nl2br turned every inter-block newline into a stray
+ * <br>, double-spacing the output. HtmlHelper::bookDescription() applies nl2br
+ * only to plain-text values and leaves block HTML untouched. The helper is
+ * intentionally specific to `libri.descrizione`; other fields are outside #313.
+ *
+ * Run:  php tests/richtext-nl2br-313.unit.php   (exit 0 iff all pass)
+ */
+
+$root = dirname(__DIR__);
+require $root . '/vendor/autoload.php';
+
+use App\Support\HtmlHelper;
+
+$pass = 0;
+$fail = 0;
+$check = static function (bool $ok, string $label) use (&$pass, &$fail): void {
+    if ($ok) { $pass++; echo "  OK  {$label}\n"; }
+    else     { $fail++; echo "  FAIL {$label}\n"; }
+};
+
+// ── 1. The bug: editor HTML must NOT gain <br> between paragraphs ────────────
+$html = "<p>Para uno.</p>\n<p>Para due.</p>\n<p>Para tre.</p>";
+$out = HtmlHelper::bookDescription($html);
+$check(!str_contains($out, '<br'), "01 HTML paragraphs render without an injected <br> (issue #313)");
+$check(substr_count($out, '<p>') === 3, "02 all three paragraphs are preserved");
+
+// ── 2. Plain text still gets its line breaks ────────────────────────────────
+$plain = "Prima riga\nSeconda riga\nTerza riga";
+$out = HtmlHelper::bookDescription($plain);
+$check(substr_count($out, '<br') === 2, "03 plain text keeps its newlines as <br>");
+
+// A single-block value that only uses <br> is already HTML → left as-is (no
+// doubling): the two <br> stay two, they are not turned into four.
+$brHtml = "Riga uno<br>Riga due";
+$out = HtmlHelper::bookDescription($brHtml);
+$check(substr_count($out, '<br') === 1, "04 a <br>-only value is treated as HTML, not re-broken");
+
+// Other block markers are detected too (div / ul / heading).
+foreach (['<div>x</div>', '<ul><li>x</li></ul>', '<h2>x</h2>'] as $i => $block) {
+    $withNl = $block . "\ntail";
+    $out = HtmlHelper::bookDescription($withNl);
+    // The trailing "\ntail" newline must not become a <br> when the value is HTML.
+    $check(!str_contains($out, '<br'), sprintf("0%d block HTML (%s) suppresses nl2br", 5 + $i, strtok($block, '>') . '>'));
+}
+
+// ── 3. Sanitisation is still applied ────────────────────────────────────────
+$out = HtmlHelper::bookDescription('<p>ok</p><script>alert(1)</script>');
+$check(!str_contains($out, '<script'), "08 script tags are still stripped (sanitised)");
+
+$out = HtmlHelper::bookDescription("plain <script>alert(1)</script> text");
+$check(!str_contains($out, '<script'), "09 script in plain-text path is stripped too");
+
+// ── 4. Empty / null ─────────────────────────────────────────────────────────
+$check(HtmlHelper::bookDescription(null) === '', "10 null => empty string");
+$check(HtmlHelper::bookDescription('') === '', "11 empty => empty string");
+
+// ── 5. Cyrillic content (the reporter's data) survives ──────────────────────
+$cyr = "<p>Америка&hellip;</p>\n<p>Однако&hellip;</p>";
+$out = HtmlHelper::bookDescription($cyr);
+$check(str_contains($out, 'Америка') && !str_contains($out, '<br'), "12 Cyrillic paragraphs render clean (reporter's case)");
+
+echo "\n{$pass} PASS, {$fail} FAIL\n";
+exit($fail === 0 ? 0 : 1);