diff --git a/.changes/unreleased/Fixed-20260708-184106.yaml b/.changes/unreleased/Fixed-20260708-184106.yaml
new file mode 100644
index 0000000..c1a68f4
--- /dev/null
+++ b/.changes/unreleased/Fixed-20260708-184106.yaml
@@ -0,0 +1,3 @@
+kind: Fixed
+body: 'logging: escape DEL, C1 controls, and invalid UTF-8 in field values'
+time: 2026-07-08T18:41:06.151703+02:00
diff --git a/log.c b/log.c
index 7d24210..7856f8e 100644
--- a/log.c
+++ b/log.c
@@ -4,6 +4,8 @@
 
 static int journal_mode = 0;
 
+static const char hexd[] = "0123456789abcdef";
+
 void
 log_setup(void)
 {
@@ -20,14 +22,56 @@ log_setup(void)
 	}
 }
 
+/* Return the length of the strict UTF-8 sequence at p (RFC 3629: no
+ * overlongs, no surrogates, max U+10FFFF) when it encodes a code point that
+ * may pass through unescaped (>= U+00A0, past the C1 controls), else 0. */
+static int
+utf8_passthrough(const unsigned char *p)
+{
+	unsigned r;
+	int len, i;
+
+	if ((p[0] & 0xe0) == 0xc0)      { len = 2; r = p[0] & 0x1f; }
+	else if ((p[0] & 0xf0) == 0xe0) { len = 3; r = p[0] & 0x0f; }
+	else if ((p[0] & 0xf8) == 0xf0) { len = 4; r = p[0] & 0x07; }
+	else
+		return 0;
+	for (i = 1; i < len; i++) {
+		if ((p[i] & 0xc0) != 0x80)
+			return 0;
+		r = (r << 6) | (p[i] & 0x3f);
+	}
+	if (len == 2 && r < 0xa0)     return 0;   /* overlong or C1 control */
+	if (len == 3 && r < 0x800)    return 0;   /* overlong */
+	if (len == 4 && r < 0x10000)  return 0;   /* overlong */
+	if (r >= 0xd800 && r <= 0xdfff)
+		return 0;
+	if (r > 0x10ffff)
+		return 0;
+	return len;
+}
+
 static int
 needs_quote(const char *s)
 {
-	if (!*s)
+	const unsigned char *p = (const unsigned char *)s;
+	int len;
+
+	if (!*p)
 		return 1;
-	for (const unsigned char *p = (const unsigned char *)s; *p; p++)
-		if (*p == ' ' || *p == '=' || *p == '"' || *p == '\\' || *p < 0x20)
+	while (*p) {
+		if (*p == ' ' || *p == '=' || *p == '"' || *p == '\\' ||
+		    *p < 0x20 || *p == 0x7f)
 			return 1;
+		if (*p < 0x80) {
+			p++;
+			continue;
+		}
+		len = utf8_passthrough(p);
+		if (!len)
+			return 1;
+		p += len;
+	}
 	return 0;
 }
 
@@ -47,19 +91,29 @@ log_enc(char *out, size_t outsz, const char *val)
 	}
 	/* quoted form */
 #define PUT(c) do { if (n + 1 < outsz) out[n++] = (c); } while (0)
+#define PUTX(c) do { PUT('\\'); PUT('x'); PUT(hexd[(c) >> 4]); PUT(hexd[(c) & 0xf]); } while (0)
 	PUT('"');
-	for (const unsigned char *p = (const unsigned char *)val; *p; p++) {
+	for (const unsigned char *p = (const unsigned char *)val; *p; ) {
 		unsigned char c = *p;
-		if (c == '"' || c == '\\') { PUT('\\'); PUT(c); }
-		else if (c == '\n') { PUT('\\'); PUT('n'); }
-		else if (c == '\t') { PUT('\\'); PUT('t'); }
-		else if (c == '\r') { PUT('\\'); PUT('r'); }
-		else if (c < 0x20) {
-			static const char hexd[] = "0123456789abcdef";
-			PUT('\\'); PUT('x'); PUT(hexd[c >> 4]); PUT(hexd[c & 0xf]);
-		} else PUT(c);
+		int len;
+
+		if (c == '"' || c == '\\') { PUT('\\'); PUT(c); p++; }
+		else if (c == '\n') { PUT('\\'); PUT('n'); p++; }
+		else if (c == '\t') { PUT('\\'); PUT('t'); p++; }
+		else if (c == '\r') { PUT('\\'); PUT('r'); p++; }
+		else if (c < 0x20 || c == 0x7f) { PUTX(c); p++; }
+		else if (c < 0x80) { PUT(c); p++; }
+		else if ((len = utf8_passthrough(p)) == 0) {
+			/* invalid UTF-8 or a C1 control: escape this
+			 * byte, retry at the next */
+			PUTX(c);
+			p++;
+		} else {
+			while (len--) { PUT(*p); p++; }
+		}
 	}
 	PUT('"');
+#undef PUTX
 #undef PUT
 	out[n < outsz ? n : outsz - 1] = '\0';
 	return n;
@@ -190,7 +244,6 @@ static char hexbuf_buf[2 * HEXBUF_MAX_IN + 1];
 const char *
 log_hexbuf(const void *buf, size_t len)
 {
-	static const char hexd[] = "0123456789abcdef";
 	const unsigned char *b = buf;
 	size_t i, n = len > HEXBUF_MAX_IN ? HEXBUF_MAX_IN : len;
 	for (i = 0; i < n; i++) {
diff --git a/t/test_log.c b/t/test_log.c
index d651e6b..2b5e8eb 100644
--- a/t/test_log.c
+++ b/t/test_log.c
@@ -8,7 +8,7 @@ is_enc(const char *val, const char *expected)
 {
 	char buf[256];
 	log_enc(buf, sizeof(buf), val);
-	if (!ok(strcmp(buf, expected) == 0, "enc(%s)", val))
+	if (!ok(strcmp(buf, expected) == 0, "enc -> %s", expected))
 		tap_diag("got: %s want: %s", buf, expected);
 }
 
@@ -37,6 +37,26 @@ main(void)
 	is_enc("t\tend",        "\"t\\tend\"");
 	is_enc("r\rend",        "\"r\\rend\"");
 	is_enc("\x01",          "\"\\x01\"");
+	/* DEL and C1 controls must not reach the output raw */
+	is_enc("\x7f",          "\"\\x7f\"");
+	is_enc("a\x7f" "b",     "\"a\\x7fb\"");
+	is_enc("\x9b",          "\"\\x9b\"");            /* bare continuation byte */
+	is_enc("\xc2\x9b",      "\"\\xc2\\x9b\"");       /* U+009B CSI */
+	is_enc("\xc2\x80",      "\"\\xc2\\x80\"");       /* U+0080 */
+	/* valid printable UTF-8 passes through, bare or quoted */
+	is_enc("bl\xc3\xa5" "b\xc3\xa6r", "bl\xc3\xa5" "b\xc3\xa6r");
+	is_enc("s\xc3\xb8 blue", "\"s\xc3\xb8 blue\"");
+	is_enc("\xc2\xa0",      "\xc2\xa0");             /* U+00A0, first past C1 */
+	is_enc("\xe2\x82\xac",  "\xe2\x82\xac");         /* U+20AC */
+	is_enc("\xf0\x9f\x98\x80", "\xf0\x9f\x98\x80");  /* U+1F600 */
+	/* malformed UTF-8 is escaped byte by byte */
+	is_enc("\xc0\x9b",      "\"\\xc0\\x9b\"");       /* overlong 2-byte */
+	is_enc("\xe0\x80\xa0",  "\"\\xe0\\x80\\xa0\"");  /* overlong 3-byte */
+	is_enc("\xed\xa0\x80",  "\"\\xed\\xa0\\x80\"");  /* surrogate U+D800 */
+	is_enc("\xf4\x90\x80\x80", "\"\\xf4\\x90\\x80\\x80\""); /* above U+10FFFF */
+	is_enc("\xc3",          "\"\\xc3\"");            /* truncated at end */
+	is_enc("\xe2\x80",      "\"\\xe2\\x80\"");       /* truncated at end */
+	is_enc("\xe2" "AB",     "\"\\xe2AB\"");          /* resync after bad lead */
 	ok(strcmp(log_u(42u), "42") == 0, "U(42)");
 	ok(strcmp(log_i(-7), "-7") == 0, "I(-7)");
 	ok(strcmp(log_hex(0xa2u), "0xa2") == 0, "HEX(0xa2)");