diff --git a/include/collection.h b/include/collection.h
index 5fa350e3..71afb27d 100644
--- a/include/collection.h
+++ b/include/collection.h
@@ -226,7 +226,7 @@ public:
                           size_t drop_tokens_threshold = Index::DROP_TOKENS_THRESHOLD,
                           const spp::sparse_hash_set<std::string> & include_fields = spp::sparse_hash_set<std::string>(),
                           const spp::sparse_hash_set<std::string> & exclude_fields = spp::sparse_hash_set<std::string>(),
-                          size_t max_facet_values=10, size_t max_hits=500,
+                          size_t max_facet_values=10, int max_hits=500,
                           const std::string & simple_facet_query = "",
                           const size_t snippet_threshold = 30,
                           const std::string & highlight_full_fields = "",
diff --git a/src/collection.cpp b/src/collection.cpp
index 01c7da07..bd077e05 100644
--- a/src/collection.cpp
+++ b/src/collection.cpp
@@ -316,7 +316,7 @@ Option<nlohmann::json> Collection::search(const std::string & query, const std::
                                   const size_t drop_tokens_threshold,
                                   const spp::sparse_hash_set<std::string> & include_fields,
                                   const spp::sparse_hash_set<std::string> & exclude_fields,
-                                  const size_t max_facet_values, const size_t max_hits,
+                                  const size_t max_facet_values, const int _max_hits,
                                   const std::string & simple_facet_query,
                                   const size_t snippet_threshold,
                                   const std::string & highlight_full_fields,
@@ -570,11 +570,13 @@ Option<nlohmann::json> Collection::search(const std::string & query, const std::
         return Option<nlohmann::json>(422, message);
     }
 
-    const size_t results_per_page = std::min(per_page, max_hits);
+    const size_t maximum_hits = (_max_hits < 0) ? std::max((size_t)100, get_num_documents()) : size_t(_max_hits);
+
+    const size_t results_per_page = std::min(per_page, maximum_hits);
     const size_t num_results = (page * results_per_page);
 
-    if(num_results > max_hits) {
-        std::string message = "Only the first " + std::to_string(max_hits) + " results are available.";
+    if(num_results > maximum_hits) {
+        std::string message = "Only the first " + std::to_string(maximum_hits) + " results are available.";
         return Option<nlohmann::json>(422, message);
     }
 
@@ -589,7 +591,7 @@ Option<nlohmann::json> Collection::search(const std::string & query, const std::
     for(Index* index: indices) {
         index->search_params = search_args(query, search_fields, filters, facets,
                                            index_to_included_ids[index_id], index_to_excluded_ids[index_id],
-                                           sort_fields_std, facet_query, num_typos, max_facet_values, max_hits,
+                                           sort_fields_std, facet_query, num_typos, max_facet_values, maximum_hits,
                                            results_per_page, page, token_order, prefix,
                                            drop_tokens_threshold, typo_tokens_threshold);
         {
diff --git a/src/core_api.cpp b/src/core_api.cpp
index 4b82786f..489b9f4e 100644
--- a/src/core_api.cpp
+++ b/src/core_api.cpp
@@ -290,6 +290,9 @@ bool get_search(http_req & req, http_res & res) {
         } else {
             req.params[MAX_HITS] = "0";
         }
+    } else if(req.params[MAX_HITS] == "all") {
+        // engine will default to size of collection
+        req.params[MAX_HITS] = "-1";
     }
 
     if(req.params.count(SNIPPET_THRESHOLD) == 0) {
diff --git a/test/collection_test.cpp b/test/collection_test.cpp
index 23269f1f..98d1c70b 100644
--- a/test/collection_test.cpp
+++ b/test/collection_test.cpp
@@ -2265,3 +2265,41 @@ TEST_F(CollectionTest, OptionalFields) {
 
     collectionManager.drop_collection("coll1");
 }
+
+TEST_F(CollectionTest, ReturnsResultsBasedOnMaxHitsParam) {
+    std::vector<std::string> facets;
+    spp::sparse_hash_set<std::string> empty;
+    nlohmann::json results = collection->search("*", query_fields, "", facets, sort_fields, 0, 100, 1,
+            FREQUENCY, false, 1000, empty, empty, 10, 12).get();
+
+    ASSERT_EQ(12, results["hits"].size());
+    ASSERT_EQ(25, results["found"].get<int>());
+
+    // should match collection size
+
+    results = collection->search("*", query_fields, "", facets, sort_fields, 0, 100, 1,
+                                 FREQUENCY, false, 1000, empty, empty, 10, -1).get();
+
+    ASSERT_EQ(25, results["hits"].size());
+    ASSERT_EQ(25, results["found"].get<int>());
+
+    // should still respect pagination
+
+    results = collection->search("*", query_fields, "", facets, sort_fields, 0, 10, 1,
+                                 FREQUENCY, false, 1000, empty, empty, 10, -1).get();
+
+    ASSERT_EQ(10, results["hits"].size());
+    ASSERT_EQ(25, results["found"].get<int>());
+
+    results = collection->search("*", query_fields, "", facets, sort_fields, 0, 10, 2,
+                                 FREQUENCY, false, 1000, empty, empty, 10, -1).get();
+
+    ASSERT_EQ(10, results["hits"].size());
+    ASSERT_EQ(25, results["found"].get<int>());
+
+    results = collection->search("*", query_fields, "", facets, sort_fields, 0, 10, 3,
+                                 FREQUENCY, false, 1000, empty, empty, 10, -1).get();
+
+    ASSERT_EQ(5, results["hits"].size());
+    ASSERT_EQ(25, results["found"].get<int>());
+}