diff --git a/colabfold/mmseqs/search.py b/colabfold/mmseqs/search.py
index 6455f4ae..c0dffcc7 100644
--- a/colabfold/mmseqs/search.py
+++ b/colabfold/mmseqs/search.py
@@ -390,7 +390,10 @@ def main():
         help="Database preload mode 0: auto, 1: fread, 2: mmap, 3: mmap+touch",
     )
     parser.add_argument(
-        "--unpack", type=int, default=1, choices=[0, 1], help="Unpack results to loose files or keep MMseqs2 databases."
+        "--unpack", type=int, default=1, choices=[0, 1], help="Unpack results to a3m text files or keep MMseqs2 databases."
+    )
+    parser.add_argument(
+        "--merge-a3m", type=int, default=1, choices=[0, 1], help="Merge unpacked a3m files into a single a3m file."
     )
     parser.add_argument(
         "--threads", type=int, default=64, help="Number of threads to use."
@@ -522,7 +525,7 @@ def main():
                 unpack=args.unpack,
             )
 
-        if args.unpack or args.af3_json:
+        if args.merge_a3m or args.af3_json:
             id = 0
             for job_number, (
                 raw_jobname,
@@ -537,22 +540,19 @@ def main():
                 for seq in query_sequences:
                     with args.base.joinpath(f"{id}.a3m").open("r") as f:
                         unpaired_msa.append(f.read())
-                    if args.af3_json:
-                        args.base.joinpath(f"{id}.a3m").unlink()
-
-                    if args.use_env_pairing:
-                        with open(args.base.joinpath(f"{id}.paired.a3m"), 'a') as file_pair:
-                            with open(args.base.joinpath(f"{id}.env.paired.a3m"), 'r') as file_pair_env:
-                                while chunk := file_pair_env.read(10 * 1024 * 1024):
-                                    file_pair.write(chunk)
-                        if args.unpack:
-                            args.base.joinpath(f"{id}.env.paired.a3m").unlink()
+                    args.base.joinpath(f"{id}.a3m").unlink()
 
                     if len(query_seqs_cardinality) > 1:
                         with args.base.joinpath(f"{id}.paired.a3m").open("r") as f:
-                            paired_msa.append(f.read())
-                    if args.unpack:
-                        args.base.joinpath(f"{id}.paired.a3m").unlink()
+                            paired = f.read()
+                        if args.merge_a3m:
+                            args.base.joinpath(f"{id}.paired.a3m").unlink()
+                        if args.use_env_pairing:
+                            with open(args.base.joinpath(f"{id}.env.paired.a3m"), 'r') as file_pair_env:
+                                paired += file_pair_env.read()
+                            if args.merge_a3m:                            
+                                args.base.joinpath(f"{id}.env.paired.a3m").unlink()
+                        paired_msa.append(paired)
                     id += 1
 
                 if args.af3_json:
@@ -560,7 +560,7 @@ def main():
                     with open(args.base.joinpath(f"{job_number}.json"), 'w') as f:
                         f.write(json.dumps(af3.content, indent=4))
 
-                if args.unpack:   
+                if args.merge_a3m:   
                     msa = msa_to_str(
                         unpaired_msa, paired_msa, query_sequences, query_seqs_cardinality
                     )
@@ -583,10 +583,34 @@ def main():
     if args.unpack:
         # rename a3m files
         for job_number, (raw_jobname, query_sequences, query_seqs_cardinality, other_molecules) in enumerate(queries_unique):
-            os.rename(
-                args.base.joinpath(f"{job_number}.a3m"),
-                args.base.joinpath(f"{safe_filename(raw_jobname)}.a3m"),
-            )
+            if args.merge_a3m:
+                os.rename(
+                    args.base.joinpath(f"{job_number}.a3m"),
+                    args.base.joinpath(f"{safe_filename(raw_jobname)}.a3m"),
+                )
+            else:
+                # Rename unpaired, paired and env paired files.
+                nseqs = len(query_seqs_cardinality)
+                for id in range(job_number * nseqs, (job_number+1) * nseqs):
+                    if nseqs > 1:
+                        os.rename(
+                            args.base.joinpath(f"{id}.a3m"),
+                            args.base.joinpath(f"{safe_filename(raw_jobname)}_{id}.a3m"),
+                        )
+                        os.rename(
+                            args.base.joinpath(f"{id}.paired.a3m"),
+                            args.base.joinpath(f"{safe_filename(raw_jobname)}_{id}.paired.a3m"),
+                        )
+                        if args.use_env_pairing:
+                            os.rename(
+                                args.base.joinpath(f"{id}.env.paired.a3m"),
+                                args.base.joinpath(f"{safe_filename(raw_jobname)}_{id}.env.paired.a3m"),
+                            )
+                    else:
+                        os.rename(
+                            args.base.joinpath(f"{id}.a3m"),
+                            args.base.joinpath(f"{safe_filename(raw_jobname)}.a3m"),
+                        )
 
         # rename m8 files
         if args.use_templates: