diff --git a/examples/write_verify.rs b/examples/write_verify.rs
new file mode 100644
index 0000000..f3a47f0
--- /dev/null
+++ b/examples/write_verify.rs
@@ -0,0 +1,88 @@
+//! An STM32L073 example that measures write and erase times
+//!
+//! The flash chip is connected to SPI1
+//!
+
+#![no_std]
+#![no_main]
+
+extern crate panic_semihosting;
+
+use cortex_m_rt::entry;
+use cortex_m_semihosting::hprintln;
+
+use hex_display::HexDisplayExt;
+use w25q::series25::Flash;
+
+use stm32l0xx_hal as hal;
+
+use crate::hal::{pac, prelude::*, rcc::Config, spi::MODE_0};
+
+#[entry]
+fn main() -> ! {
+    hprintln!("START");
+    let p = pac::Peripherals::take().unwrap();
+    let cp = cortex_m::Peripherals::take().unwrap();
+
+    let mut rcc = p.RCC.freeze(Config::hsi16());
+
+    let gpiob = p.GPIOB.split(&mut rcc);
+
+    let sck = gpiob.pb3;
+    let miso = gpiob.pb4;
+    let mosi = gpiob.pb5;
+    let cs = gpiob.pb6.into_push_pull_output();
+
+    let spi = p
+        .SPI1
+        .spi((sck, miso, mosi), MODE_0, 4_000_000.Hz(), &mut rcc);
+
+    let mut flash = Flash::init(spi, cs).unwrap();
+    let id = flash.read_jedec_id().unwrap();
+    hprintln!("{:?}", id);
+    let mut fail_count = 0u32;
+
+    const ITERATIONS: usize = 100;
+    for i in 0..ITERATIONS {
+        let mut inb = [0u8; 512];
+        let mut inb_copy = [0u8; 512];
+        let mut outb = [0u8; 512];
+        for (n, b) in inb.iter_mut().enumerate() {
+            *b = (n + i) as u8;
+        }
+        inb_copy.copy_from_slice(&inb);
+        const ADDR: u32 = 0x0000_0000;
+        match i {
+            0..=90 => {
+                hprintln!("sector erase");
+                flash.erase_sectors(ADDR, 1).unwrap();
+            }
+            91..=98 => {
+                hprintln!("block erase");
+                flash.erase_block(ADDR).unwrap();
+            },
+            99..=100 => {
+                hprintln!("chip erase");
+                flash.erase_all().unwrap();
+            },
+            _ => (),
+        }
+        // inb will get overwritten below!
+        flash.write_bytes(ADDR, &mut inb).unwrap();
+        flash.read(ADDR, &mut outb).unwrap();
+        if outb != inb_copy {
+            hprintln!("Failed verification");
+            // hprintln!("wrote: {}", inb_copy.hex());
+            // hprintln!("read:  {}", outb.hex());
+            fail_count += 1;
+        }
+        hprintln!("write iteration {}/{}", i, ITERATIONS);
+    }
+
+    if fail_count > 0 {
+        hprintln!("num failures: {}", fail_count);
+    }
+    hprintln!("DONE");
+
+    loop {}
+}
diff --git a/src/series25.rs b/src/series25.rs
index e7cd5b0..30860f5 100644
--- a/src/series25.rs
+++ b/src/series25.rs
@@ -233,14 +233,29 @@ impl<SPI: Transfer<u8>, CS: OutputPin> Flash<SPI, CS> {
     }
 
     pub fn write_enable(&mut self) -> Result<(), Error<SPI, CS>> {
-        let mut cmd_buf = [Opcode::WriteEnable as u8];
-        self.command(&mut cmd_buf)?;
+        while !self.read_status()?.contains(Status::WEL) {
+            let mut cmd_buf = [Opcode::WriteEnable as u8];
+            self.command(&mut cmd_buf)?;
+        }
         Ok(())
     }
 
+    // Wait for the BUSY flag to clear
+    // This function also handles the cases where the BUSY
+    // flag is not set yet (but expected) as well as when the
+    // BUSY flag is briefly unset to be set again shortly after.
+    // The latter behavior has been observed on the very popular
+    // Winbond chip W25Q128JV when executing a chip erase operation.
     fn wait_done(&mut self) -> Result<(), Error<SPI, CS>> {
         // TODO: Consider changing this to a delay based pattern
-        while self.read_status()?.contains(Status::BUSY) {}
+        let mut busy = true;
+        const RETRIES :usize = 10;
+        while busy {
+            busy = false;
+            for _ in 0..RETRIES {
+                busy |= self.read_status()?.contains(Status::BUSY);
+            }
+        }
         Ok(())
     }