diff --git a/NEWS.md b/NEWS.md
index 2f38cf6..6128c3c 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -1,3 +1,14 @@
+# monolix2rx 0.0.7
+
+* Fix implicit `ptrdiff_t` to `int` truncation in `rc_dup_str`
+  (`src/shared.c`).  When the parser passes a string segment longer than
+  `INT_MAX` bytes (or a NUL-terminated string of that length), the
+  pointer difference / `strlen` result was silently cast to `int`,
+  truncating the length to a wrong (often negative) value.  The new
+  guard rejects such inputs with an informative R error.  Also adds a
+  thread-safety comment documenting that the parser globals are
+  intentionally not mutex-protected (R is single-threaded).
+
 # monolix2rx 0.0.6
 
 * Updated to add types for rstudio completion
diff --git a/src/shared.c b/src/shared.c
index d7e48d1..7699aab 100644
--- a/src/shared.c
+++ b/src/shared.c
@@ -19,6 +19,10 @@ dparserPtrIni
 #include "parseSyntaxErrors.h"
 
 // These are the shared variables
+// NOTE: These globals are intentionally not mutex-protected.
+// R's interpreter is single-threaded; its memory allocator (R_Calloc, R_Free)
+// and error handling (Rf_error) are not safe to call from multiple threads.
+// All parse operations must occur on the R main thread.
 
 const char *record;
 int _rxode2_reallyHasAfter = 0;
@@ -41,8 +45,20 @@ int lastStrLoc=0;
 vLines _dupStrs;
 char * rc_dup_str(const char *s, const char *e) {
   lastStr=s;
-  int l = e ? e-s : (int)strlen(s);
-  //syntaxErrorExtra=min(l-1, 40);
+  int l;
+  if (e) {
+    ptrdiff_t diff = e - s;
+    if (diff < 0 || diff > (ptrdiff_t)INT_MAX) {
+      Rf_error(_("string segment too long in rc_dup_str"));
+    }
+    l = (int)diff;
+  } else {
+    size_t sLen = strlen(s);
+    if (sLen > (size_t)INT_MAX) {
+      Rf_error(_("string too long in rc_dup_str"));
+    }
+    l = (int)sLen;
+  }
   addLine(&_dupStrs, "%.*s", l, s);
   return _dupStrs.line[_dupStrs.n-1];
 }
diff --git a/tests/testthat/test-mem-rc-dup-str.R b/tests/testthat/test-mem-rc-dup-str.R
new file mode 100644
index 0000000..ce8a779
--- /dev/null
+++ b/tests/testthat/test-mem-rc-dup-str.R
@@ -0,0 +1,44 @@
+test_that("rc_dup_str handles normal-sized inputs without error", {
+  # Sanity check: regular Mlxtran fragments must continue to parse cleanly
+  # after the INT_MAX guards added to rc_dup_str.
+  expect_no_error(
+    tryCatch(
+      .Call(`_monolix2rx_trans_equation`,
+            "[LONGITUDINAL] EQUATION:\nf = exp(-k*t)\n",
+            "[LONGITUDINAL] EQUATION:"),
+      error = function(e) {
+        if (grepl("rc_dup_str", conditionMessage(e))) stop(e)
+        # Other parse errors (e.g., from synthetic test input) are fine.
+        NULL
+      }
+    )
+  )
+})
+
+test_that("rc_dup_str int truncation guard triggers on huge inputs (skipped: requires ~2GB RAM)", {
+  skip("Requires ~2GB free RAM to construct a >INT_MAX-byte string and exercise the rc_dup_str path")
+  # --- What this test checks ---
+  # `rc_dup_str` (src/shared.c) computes the length as
+  #   int l = e ? e-s : (int)strlen(s);
+  # When the source string segment is larger than INT_MAX bytes, the
+  # `(int)` cast silently truncates to a wrong value, which propagates
+  # into `addLine(&_dupStrs, "%.*s", l, s)` and either reads past the
+  # buffer (OOB) or wraps to a tiny copy.
+  #
+  # The guard added by this fix range-checks the ptrdiff_t / size_t
+  # length and raises an R error before the truncation can happen.
+  #
+  # --- How to run manually (outside devtools::test()) ---
+  # Start a fresh R session with at least 3 GB of available RAM, then:
+  #
+  #   library(monolix2rx)
+  #   big <- strrep("a", 2147483647L)   # exactly INT_MAX bytes
+  #   # Before fix: silent truncation; potential heap corruption
+  #   # After fix:  error "string too long in rc_dup_str"
+
+  big <- strrep("a", 2147483647L)
+  expect_error(
+    .Call(`_monolix2rx_trans_equation`, big, "[LONGITUDINAL]"),
+    "rc_dup_str|too long"
+  )
+})