diff --git a/Cargo.toml b/Cargo.toml
index f43cf071..6b863bce 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -12,11 +12,16 @@ license = "MIT/Apache-2.0"
 
 [dependencies]
 byteorder = "1.1.0"
-safemem = "0.2.0"
+safemem = { version = "0.3.0", default-features = false }
 
 [dev-dependencies]
 rand = "0.4"
 
+[features]
+std = ["safemem/std"]
+alloc = []
+default = ["std"]
+
 [profile.bench]
 # Useful for better disassembly when using `perf record` and `perf annotate`
 debug = true
diff --git a/README.md b/README.md
index d7694dc7..5372ec40 100644
--- a/README.md
+++ b/README.md
@@ -97,6 +97,18 @@ cargo +nightly fuzz run roundtrip_mime -- -max_len=10240
 cargo +nightly fuzz run roundtrip_random_config -- -max_len=10240
 ```
 
+no_std support
+---
+
+This crate has a feature flag `std` that is enabled by default.
+
+To use in a *no_std* environment, you can disable default features. To
+use the full API, enable feature `alloc` and setup an allocator.
+
+In environments without an allocator, the API provides only the
+`encode_config_slice()` and `decode_config_slice()` functions (which
+do not use the dynamically-sized `Vec` type) and the
+`strip_whitespace` config must be disabled.
 
 License
 ---
diff --git a/src/chunked_encoder.rs b/src/chunked_encoder.rs
index e472b073..eb7e5ba6 100644
--- a/src/chunked_encoder.rs
+++ b/src/chunked_encoder.rs
@@ -1,6 +1,6 @@
 use encode::{add_padding, encode_to_slice};
 use line_wrap::line_wrap;
-use std::cmp;
+use core::cmp;
 use {Config, LineEnding, LineWrap};
 
 /// The output mechanism for ChunkedEncoder's encoded bytes.
@@ -171,6 +171,7 @@ pub mod tests {
     use tests::random_config;
     use *;
 
+    use std::prelude::v1::*;
     use std::str;
 
     use self::rand::distributions::{IndependentSample, Range};
diff --git a/src/decode.rs b/src/decode.rs
index e25b752a..c59d2141 100644
--- a/src/decode.rs
+++ b/src/decode.rs
@@ -1,7 +1,13 @@
+#[cfg(feature = "std")]
+use std::vec::Vec;
+#[cfg(feature = "alloc")]
+use alloc::Vec;
 use byteorder::{BigEndian, ByteOrder};
-use {tables, CharacterSet, Config, STANDARD};
+use {tables, CharacterSet, Config};
 
-use std::{error, fmt, str};
+use core::fmt;
+#[cfg(feature = "std")]
+use std::{error, str};
 
 // decode logic operates on chunks of 8 input bytes without padding
 const INPUT_CHUNK_LEN: usize = 8;
@@ -25,6 +31,10 @@ pub enum DecodeError {
     InvalidByte(usize, u8),
     /// The length of the input is invalid.
     InvalidLength,
+    /// This will be returned if you configure the decoder to strip
+    /// whitespace but there is not `Vec` in absence of the `alloc`
+    /// feature.
+    CannotStripWhitespace,
 }
 
 impl fmt::Display for DecodeError {
@@ -34,15 +44,18 @@ impl fmt::Display for DecodeError {
                 write!(f, "Invalid byte {}, offset {}.", byte, index)
             }
             DecodeError::InvalidLength => write!(f, "Encoded text cannot have a 6-bit remainder."),
+            DecodeError::CannotStripWhitespace => write!(f, "Configured to strip white-space but no allocator available"),
         }
     }
 }
 
+#[cfg(feature = "std")]
 impl error::Error for DecodeError {
     fn description(&self) -> &str {
         match *self {
             DecodeError::InvalidByte(_, _) => "invalid byte",
             DecodeError::InvalidLength => "invalid length",
+            DecodeError::CannotStripWhitespace => "cannot strip whitespace",
         }
     }
 
@@ -65,7 +78,9 @@ impl error::Error for DecodeError {
 ///    println!("{:?}", bytes);
 ///}
 ///```
+#[cfg(any(feature = "std", feature = "alloc"))]
 pub fn decode<T: ?Sized + AsRef<[u8]>>(input: &T) -> Result<Vec<u8>, DecodeError> {
+    use STANDARD;
     decode_config(input, STANDARD)
 }
 
@@ -85,6 +100,7 @@ pub fn decode<T: ?Sized + AsRef<[u8]>>(input: &T) -> Result<Vec<u8>, DecodeError
 ///    println!("{:?}", bytes_url);
 ///}
 ///```
+#[cfg(any(feature = "std", feature = "alloc"))]
 pub fn decode_config<T: ?Sized + AsRef<[u8]>>(
     input: &T,
     config: Config,
@@ -115,6 +131,7 @@ pub fn decode_config<T: ?Sized + AsRef<[u8]>>(
 ///    println!("{:?}", buffer);
 ///}
 ///```
+#[cfg(any(feature = "std", feature = "alloc"))]
 pub fn decode_config_buf<T: ?Sized + AsRef<[u8]>>(
     input: &T,
     config: Config,
@@ -162,10 +179,17 @@ pub fn decode_config_slice<T: ?Sized + AsRef<[u8]>>(
     config: Config,
     output: &mut [u8],
 ) -> Result<usize, DecodeError> {
+    #[cfg(any(feature = "std", feature = "alloc"))]
     let input_copy;
     let input_bytes = if config.strip_whitespace {
-        input_copy = copy_without_whitespace(input.as_ref());
-        input_copy.as_ref()
+        #[cfg(not(any(feature = "std", feature = "alloc")))]
+        return Err(DecodeError::CannotStripWhitespace);
+
+        #[cfg(any(feature = "std", feature = "alloc"))]
+        {
+            input_copy = copy_without_whitespace(input.as_ref());
+            input_copy.as_ref()
+        }
     } else {
         input.as_ref()
     };
@@ -186,6 +210,7 @@ fn num_chunks(input: &[u8]) -> usize {
         .expect("Overflow when calculating number of chunks in input") / INPUT_CHUNK_LEN
 }
 
+#[cfg(any(feature = "std", feature = "alloc"))]
 fn copy_without_whitespace(input: &[u8]) -> Vec<u8> {
     let mut input_copy = Vec::<u8>::with_capacity(input.len());
     input_copy.extend(input.iter().filter(|b| !b" \n\t\r\x0b\x0c".contains(b)));
@@ -540,6 +565,7 @@ mod tests {
 
     use self::rand::distributions::{IndependentSample, Range};
     use self::rand::Rng;
+    use std::prelude::v1::*;
 
     #[test]
     fn decode_chunk_precise_writes_only_6_bytes() {
diff --git a/src/display.rs b/src/display.rs
index 53ee1d7b..63d64fb9 100644
--- a/src/display.rs
+++ b/src/display.rs
@@ -11,8 +11,8 @@
 
 use super::chunked_encoder::{ChunkedEncoder, ChunkedEncoderError};
 use super::Config;
-use std::fmt::{Display, Formatter};
-use std::{fmt, str};
+use core::fmt::{Display, Formatter};
+use core::{fmt, str};
 
 // I'm not convinced that we should expose ChunkedEncoder or its error type since it's just an
 // implementation detail, so use a different error type.
@@ -82,6 +82,7 @@ mod tests {
                                                SinkTestHelper};
     use super::super::*;
     use super::*;
+    use std::string::String;
 
     #[test]
     fn basic_display() {
diff --git a/src/encode.rs b/src/encode.rs
index be1eb86a..ca4a9d65 100644
--- a/src/encode.rs
+++ b/src/encode.rs
@@ -1,5 +1,9 @@
+#[cfg(feature = "std")]
+use std::string::String;
+#[cfg(feature = "alloc")]
+use alloc::String;
 use byteorder::{BigEndian, ByteOrder};
-use {line_wrap, line_wrap_parameters, Config, LineWrap, STANDARD};
+use {line_wrap, line_wrap_parameters, Config, LineWrap};
 
 ///Encode arbitrary octets as base64.
 ///Returns a String.
@@ -15,7 +19,9 @@ use {line_wrap, line_wrap_parameters, Config, LineWrap, STANDARD};
 ///    println!("{}", b64);
 ///}
 ///```
+#[cfg(any(feature = "std", feature = "alloc"))]
 pub fn encode<T: ?Sized + AsRef<[u8]>>(input: &T) -> String {
+    use STANDARD;
     encode_config(input, STANDARD)
 }
 
@@ -35,6 +41,7 @@ pub fn encode<T: ?Sized + AsRef<[u8]>>(input: &T) -> String {
 ///    println!("{}", b64_url);
 ///}
 ///```
+#[cfg(any(feature = "std", feature = "alloc"))]
 pub fn encode_config<T: ?Sized + AsRef<[u8]>>(input: &T, config: Config) -> String {
     let mut buf = match encoded_size(input.as_ref().len(), &config) {
         Some(n) => String::with_capacity(n),
@@ -64,6 +71,7 @@ pub fn encode_config<T: ?Sized + AsRef<[u8]>>(input: &T, config: Config) -> Stri
 ///    println!("{}", buf);
 ///}
 ///```
+#[cfg(any(feature = "std", feature = "alloc"))]
 pub fn encode_config_buf<T: ?Sized + AsRef<[u8]>>(input: &T, config: Config, buf: &mut String) {
     let input_bytes = input.as_ref();
 
@@ -354,6 +362,7 @@ mod tests {
     use self::rand::distributions::{IndependentSample, Range};
     use self::rand::Rng;
     use std;
+    use std::prelude::v1::*;
     use std::str;
 
     #[test]
diff --git a/src/lib.rs b/src/lib.rs
index 020592fc..e730c2e4 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -60,6 +60,15 @@
     unused_results, variant_size_differences, warnings
 )]
 
+#![cfg_attr(feature = "alloc", feature(alloc))]
+#![no_std]
+
+#[cfg(feature = "std")]
+#[macro_use]
+extern crate std;
+#[cfg(feature = "alloc")]
+extern crate alloc;
+
 extern crate byteorder;
 
 mod chunked_encoder;
@@ -70,10 +79,14 @@ mod tables;
 use line_wrap::{line_wrap, line_wrap_parameters};
 
 mod encode;
-pub use encode::{encode, encode_config, encode_config_buf, encode_config_slice};
+pub use encode::encode_config_slice;
+#[cfg(any(feature = "std", feature = "alloc"))]
+pub use encode::{encode, encode_config, encode_config_buf};
 
 mod decode;
-pub use decode::{decode, decode_config, decode_config_buf, decode_config_slice, DecodeError};
+pub use decode::{decode_config_slice, DecodeError};
+#[cfg(any(feature = "std", feature = "alloc"))]
+pub use decode::{decode, decode_config, decode_config_buf};
 
 #[cfg(test)]
 mod tests;
diff --git a/src/line_wrap.rs b/src/line_wrap.rs
index ca81be96..80be5987 100644
--- a/src/line_wrap.rs
+++ b/src/line_wrap.rs
@@ -165,6 +165,7 @@ mod tests {
 
     use self::rand::distributions::{IndependentSample, Range};
     use self::rand::Rng;
+    use std::prelude::v1::*;
 
     #[test]
     fn line_params_perfect_multiple_of_line_length_lf() {
diff --git a/src/tests.rs b/src/tests.rs
index ff0f4b67..5c50f5f8 100644
--- a/src/tests.rs
+++ b/src/tests.rs
@@ -4,6 +4,7 @@ use encode::encoded_size;
 use line_wrap::line_wrap_parameters;
 use *;
 
+use std::prelude::v1::*;
 use std::str;
 
 use self::rand::distributions::{IndependentSample, Range};